<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:wfw="http://wellformedweb.org/CommentAPI/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	xmlns:atom="http://www.w3.org/2005/Atom"
	xmlns:sy="http://purl.org/rss/1.0/modules/syndication/"
	xmlns:slash="http://purl.org/rss/1.0/modules/slash/" > 

	<channel>
		<title>Bill Hartzer</title>
		<atom:link href="https://www.billhartzer.com/feed/gn" rel="self" type="application/rss+xml" />
		<link>https://www.billhartzer.com</link>
		<description></description>
		<lastBuildDate>Wed, 14 Oct 2026 23:53:20 +0000</lastBuildDate>
		<language>en-US</language>
		<sy:updatePeriod> 
			hourly 
		</sy:updatePeriod>
		<sy:updateFrequency> 
			1 
		</sy:updateFrequency>
		<atom:link rel="hub" href="https://pubsubhubbub.appspot.com/" />
		<generator>GN Publisher v1.5.30 https://wordpress.org/plugins/gn-publisher/</generator>

			<item>
				<title>Why Google Says 77% of My SearchEngines.net Traffic Is Branded</title>
				<link>https://www.billhartzer.com/seo/google-search-console-branded-traffic-generic-domain/</link>
				<pubDate>Wed, 14 Oct 2026 23:53:20 +0000</pubDate>
								<dc:creator><![CDATA[Bill Hartzer]]></dc:creator>				<guid isPermaLink="false">https://www.billhartzer.com/?p=18538</guid>
					<description><![CDATA[Here&#8217;s something interesting about owning a generic, keyword-rich domain name: My website, SearchEngines.net, gets most...]]></description>

				<content:encoded><![CDATA[<p><img fetchpriority="high" decoding="async" src="https://www.billhartzer.com/wp-content/uploads/2026/10/Screenshot-2026-10-13-at-12.11.42-PM.png" alt="" width="692" height="297" class="alignnone size-full wp-image-18539" srcset="https://www.billhartzer.com/wp-content/uploads/2026/10/Screenshot-2026-10-13-at-12.11.42-PM.png 692w, https://www.billhartzer.com/wp-content/uploads/2026/10/Screenshot-2026-10-13-at-12.11.42-PM-300x129.png 300w" sizes="(max-width: 692px) 100vw, 692px" /></p>
<p>Here&#8217;s something interesting about owning a generic, keyword-rich domain name: My website, <a href="https://SearchEngines.net">SearchEngines.net</a>, gets most of its Google search clicks from queries that Google Search Console classifies as branded.</p>
<p><span id="more-18538"></span></p>
<p>In fact, 77% of my clicks are classified as branded traffic, compared with just 23% classified as non-branded.</p>
<p>But when I look at the actual search queries driving traffic to the site, something doesn&#8217;t quite add up.</p>
<p>Some of the top queries include:</p>
<ul>
<li>search engines</li>
<li>search engine</li>
<li>best search engine</li>
<li>web search engine</li>
<li>all search engines</li>
<li>search engine directory</li>
<li>directory of search engines</li>
<li>most used search engines in the world</li>
</ul>
<p>These are generic search phrases. People searching for them could be looking for information, a directory, a list of search engines, or a website that helps them find search tools.</p>
<p>They aren&#8217;t necessarily searching for a particular company.</p>
<p>Of course, some queries, such as &#8220;searchengines&#8221; and &#8220;search engine net,&#8221; could indicate that someone is specifically looking for my website. That&#8217;s a reasonable interpretation of those searches.</p>
<p>But what about searches for &#8220;search engine,&#8221; &#8220;best search engine,&#8221; or &#8220;web search engine&#8221;?</p>
<p>Those phrases describe a category, not necessarily a brand.</p>
<p>This raises an interesting question about how Google interprets branded search traffic when a website&#8217;s domain name is also a generic keyword.</p>
<h2>What Google&#8217;s Branded and Non-Branded Filter Actually Does</h2>
<p>Google introduced a branded queries filter in Google Search Console in November 2025. The feature became available to all eligible sites in March 2026.</p>
<p>The filter allows website owners to separate search performance data into two categories: branded and non-branded queries.</p>
<p>According to <a href="https://developers.google.com/search/blog/2025/11/search-console-branded-filter">Google&#8217;s Search Central documentation</a>, branded queries can include a brand name, common misspellings and variations, and searches associated with a brand&#8217;s specific products or services.</p>
<p>Non-branded queries generally include searches that don&#8217;t indicate a specific brand.</p>
<p>The purpose is straightforward. If someone searches for a company&#8217;s name, that search may indicate existing brand awareness. If someone searches for a general service or product category, that search may represent an opportunity to reach a new audience.</p>
<p>For example, someone searching for &#8220;Acme Plumbing&#8221; is likely looking for a particular business. Someone searching for &#8220;plumber near me&#8221; may be looking for a plumbing company without having one in mind.</p>
<p>That distinction can be useful when measuring brand awareness and evaluating SEO performance.</p>
<p>Google&#8217;s implementation, however, has an important detail: It doesn&#8217;t rely solely on a list of keywords or a simple rule that checks whether a query contains a brand name. Google says it uses an internal, AI-assisted system to classify queries, including some that don&#8217;t explicitly contain the brand name.</p>
<p>Google also acknowledges that some queries may be misclassified. The feature is intended to help analyze search performance, and its classifications do not affect search rankings.</p>
<p>That last point is important. The branded classification is a reporting feature, not a ranking signal.</p>
<h2>Why SearchEngines.net Is an Interesting Example</h2>
<p>My website presents an unusual situation because the domain name itself consists of generic keywords.</p>
<p>The words &#8220;search engines&#8221; describe an entire category of websites and services. They&#8217;re also the primary words people use when looking for information about search engines.</p>
<p>The domain name, SearchEngines.net, matches those words almost exactly.</p>
<p>This creates a potential overlap between two different types of search intent.</p>
<p><strong>Brand-oriented intent:</strong> Someone types &#8220;SearchEngines.net&#8221; because they know about the site and want to visit it.</p>
<p><strong>Generic search intent:</strong> Someone types &#8220;search engines&#8221; because they want to find a list of search engines, compare search tools, or discover a search engine directory.</p>
<p>Both searches can lead to the same website. However, the person making the second search may have no prior knowledge of the website or its name.</p>
<p>When Google classifies a large percentage of my traffic as branded, it raises the possibility that the generic nature of the domain is affecting how those searches are interpreted.</p>
<p>I don&#8217;t have evidence that Google is classifying every generic query as branded simply because the domain contains those keywords. The aggregate percentages alone don&#8217;t establish that.</p>
<p>But the combination of a 77% branded traffic share and a list of generic top search queries makes this worth investigating.</p>
<h2>Does a Keyword-Rich Domain Change the Meaning of Branded Traffic?</h2>
<p>Consider the difference between two hypothetical websites.</p>
<p>The first is a company called Acme Search Technologies, operating on a company-name domain. The second is SearchEngines.net, which uses a generic phrase as its domain name.</p>
<p>The company-name website has a relatively clear distinction between branded and non-branded searches. A search for its company name is likely branded, while a search for &#8220;best search engine&#8221; is more likely to be non-branded.</p>
<p>SearchEngines.net is different.</p>
<p>A search for &#8220;search engines&#8221; could be a generic search for the category. It could also be a search for a website whose name closely matches that phrase.</p>
<p>The words alone don&#8217;t tell us which interpretation is correct.</p>
<p>This is where domain names can complicate the way we interpret SEO reports. A generic domain name can serve as both a descriptive keyword phrase and the name of a specific website.</p>
<p>That doesn&#8217;t mean every generic domain will receive the same classification. Google&#8217;s system may consider other signals and contextual information when identifying a brand.</p>
<p>It does mean that website owners should be careful about treating the branded traffic percentage as a direct measure of how many people recognize their brand.</p>
<p>A high branded traffic share may indicate brand recognition, but the percentage alone doesn&#8217;t prove it.</p>
<h2>How This Affects SEO Reporting</h2>
<p>The distinction between branded and non-branded traffic matters because SEO professionals often use these metrics to measure growth.</p>
<p>For a business with an established name, an increase in branded searches may indicate that more people recognize the company, remember its name, or are returning to its website.</p>
<p>An increase in non-branded searches may indicate improved visibility for general product or service queries and opportunities to reach new audiences.</p>
<p>However, those interpretations depend on the nature of the business, its domain name, and the queries being classified.</p>
<p>For a website with a generic, keyword-rich domain, the same percentages may tell a different story.</p>
<p>If Google Search Console reports that 77% of clicks are branded, it would be easy to conclude that most visitors already know the website. But if many of those clicks come from generic searches, that conclusion may be premature.</p>
<p>I&#8217;d want to examine the actual queries, compare them with the site&#8217;s name and brand variations, and determine whether the branded classification makes sense for each query.</p>
<p>It&#8217;s also worth comparing the branded and non-branded reports over time. A change in the percentages could reflect a change in search behavior, traffic composition, or Google&#8217;s classification—not necessarily a corresponding change in brand awareness.</p>
<p>In other words, the classification is useful data, but it needs context.</p>
<h2>How to Investigate This in Google Search Console</h2>
<p>If you own a website with a generic domain name, you can investigate the issue using Google Search Console.</p>
<p>Start with the Search results Performance report and select the branded or non-branded query filter. You can compare clicks, impressions, click-through rate, and average position for each group.</p>
<p>Then examine the actual queries associated with the site.</p>
<p>Look for three types of searches:</p>
<ol>
<li><strong>Clear brand searches.</strong> These include the exact domain name, distinctive brand terms, and common variations that strongly suggest someone is looking for your website.</li>
<li><strong>Generic category searches.</strong> These include broad phrases describing a product, service, industry, or topic.</li>
<li><strong>Ambiguous searches.</strong> These are phrases that could refer either to your website or to the broader category in which it operates.</li>
</ol>
<p>For a site like SearchEngines.net, the third group is especially interesting. Even a phrase such as &#8220;search engines&#8221; can be difficult to interpret without knowing the searcher&#8217;s intent.</p>
<p>You should also keep Google&#8217;s reporting limitations in mind. The branded filter is available only for eligible properties with sufficient search data, and its historical coverage begins in March 2025. Google notes that some queries may be classified incorrectly.</p>
<p>The goal isn&#8217;t to assume Google is wrong. It&#8217;s to understand what the classification tells you—and what it doesn&#8217;t.</p>
<h2>What This Means for Generic Domain Names</h2>
<p>There is another question worth considering here: Does owning a generic, keyword-rich domain name change the way we should evaluate its SEO performance?</p>
<p>A company using a new, invented brand name generally has to establish recognition for that name while also competing for generic search terms relevant to its business.</p>
<p>A domain such as SearchEngines.net starts from a different position. Its words already correspond to a recognizable topic and the language people use to search for that topic.</p>
<p>That doesn&#8217;t guarantee rankings, clicks, or a particular SEO advantage. The site&#8217;s content, relevance, competition, authority, and many other factors still matter.</p>
<p>But it does create an unusual measurement problem. The same words can describe the website&#8217;s brand and the subject matter people are searching for.</p>
<p>As a result, branded versus non-branded reporting may be harder to interpret than it is for a website with a distinctive company name.</p>
<p>For domain investors, marketers, and SEO professionals, that&#8217;s worth remembering when evaluating traffic data or comparing websites with different types of domain names.</p>
<p>A generic domain may have a name that matches the search query, but that doesn&#8217;t necessarily mean the searcher recognizes the domain as a brand.</p>
<h2>The Question I&#8217;m Still Investigating</h2>
<p>My SearchEngines.net data raises a question I think is worth exploring further.</p>
<p>If Google Search Console classifies 77% of my clicks as branded, but many of the site&#8217;s top queries are generic phrases such as &#8220;search engine,&#8221; &#8220;best search engine,&#8221; and &#8220;web search engine,&#8221; how much of that traffic represents genuine brand recognition?</p>
<p>And how much might reflect the overlap between a generic keyword and a domain name?</p>
<p>The available data doesn&#8217;t answer that question by itself. To establish what&#8217;s happening, I&#8217;d need to examine the query-level classifications and compare them with the site&#8217;s actual brand-related searches.</p>
<p>Still, the example illustrates why SEO reporting requires more than looking at percentages in a dashboard.</p>
<p><strong>When your domain name is also a generic keyword, the line between a branded search and a non-branded search can get blurry.</strong></p>
<p>I&#8217;d be interested to hear from other website owners and SEO professionals: Have you seen similar results for a generic, keyword-rich domain name? Does Google Search Console classify a surprisingly large share of your traffic as branded, even when the actual queries look generic?</p>
]]></content:encoded>
					</item>
	
			<item>
				<title>Paid Search Gets a Second Act: fullthrottle.ai Turns Search and Social Clicks Into Household Audiences</title>
				<link>https://www.billhartzer.com/search-engine-marketing/paid-search-fullthrottle-ai-search-social-clicks/</link>
				<pubDate>Tue, 06 Oct 2026 22:41:46 +0000</pubDate>
								<dc:creator><![CDATA[Bill Hartzer]]></dc:creator>				<guid isPermaLink="false">https://www.billhartzer.com/?p=18535</guid>
					<description><![CDATA[Paid search and paid social can tell marketers a lot about what people want. A...]]></description>

				<content:encoded><![CDATA[<p>Paid search and paid social can tell marketers a lot about what people want. A search, a click, or an ad engagement can be a strong signal of interest.</p>
<p><span id="more-18535"></span></p>
<p>But what happens after that signal?</p>
<p>That is the problem fullthrottle.ai is taking aim at with its new Last Mile Audiences capability. The Philadelphia-based AdTech company announced the launch on Oct. 6, 2026, giving marketers a way to turn engagement from paid search and paid social into addressable household audiences that can be reached through additional media channels.</p>
<p>The idea is fairly simple: Don&#8217;t let a valuable click disappear inside Google, Meta, or another advertising platform.</p>
<p>Instead, fullthrottle.ai can identify households that have already shown interest and make those households available for additional campaigns across connected TV, online video, audio, and direct mail.</p>
<h2>Turning a Click Into Something Marketers Can Keep Using</h2>
<p>Paid search and paid social are often treated as performance marketing channels. The goal is usually clear. Get someone to click, generate a lead, make a sale, or produce another measurable action.</p>
<p>The problem is that consumers rarely follow such a neat path.</p>
<p>Someone might search for a service, click an advertisement, visit a website, and then leave. Another person might engage with a social ad but decide that today is not the day to make a purchase.</p>
<p>The advertising investment has still produced something valuable: a signal of consumer intent.</p>
<p>Last Mile Audiences is built around making use of that signal after the original interaction. fullthrottle.ai says marketers can transform those engagement signals into first-party, addressable household audiences inside its platform.</p>
<p>That creates a different way to look at paid media. A click does not have to be the end of the process. It can become the beginning of the next phase.</p>
<h2>From Search and Social to CTV, Video, Audio and Direct Mail</h2>
<p>Once an audience has been identified, marketers can activate those households across several channels through the fullthrottle.ai platform.</p>
<p>That includes:</p>
<ul>
<li>Connected TV (CTV)</li>
<li>Online video</li>
<li>Audio</li>
<li>Display</li>
<li>Direct mail</li>
</ul>
<p>This matters because consumers do not spend their entire day inside one advertising platform.</p>
<p>A person might search Google for a home service in the morning, watch television that evening, listen to a podcast the next day, and receive a piece of direct mail later in the week. Those interactions can look like separate events when viewed through individual advertising systems.</p>
<p>fullthrottle.ai is taking a broader view. The same household that showed intent through search or social can become an audience for additional media.</p>
<p>That gives marketers another opportunity to continue the conversation instead of starting from scratch.</p>
<h2>Recency Becomes Part of the Audience Strategy</h2>
<p>One of the more interesting parts of the announcement is the addition of rolling household windows.</p>
<p>fullthrottle.ai will segment households according to how recently they showed activity. The four windows are:</p>
<ul>
<li>0–30 days</li>
<li>31–60 days</li>
<li>61–90 days</li>
<li>91–120 days</li>
</ul>
<p>That may sound like a small feature. It is not.</p>
<p>Recency can make a major difference in advertising. Someone who searched for a plumber yesterday is probably a different marketing opportunity from someone who searched for plumbing services three months ago.</p>
<p>The same principle applies to larger purchases. A consumer researching a vehicle, home improvement project, financial service, or other high-consideration product may take weeks or months before making a decision.</p>
<p>Marketers can select individual audience windows or combine several of them based on the expected buying cycle.</p>
<h3>Home Services Provide a Straightforward Example</h3>
<p>Consider a home services company.</p>
<p>A household that searched for an HVAC repair company seven days ago may deserve a different message from a household that showed similar interest 100 days ago.</p>
<p>The first household could still be actively looking for a provider. The second may have already hired someone, abandoned the project, or simply moved on.</p>
<p>By separating audiences according to recency, marketers have another lever for deciding who should receive a message and when.</p>
<p>That is a practical distinction. Good advertising is often about reaching the right people at the right point in the buying process, not simply reaching more people.</p>
<h2>AI Can Recommend the Audiences, Too</h2>
<p>Marketers will not have to create a separate audience-building process every time they want to use Last Mile Audiences.</p>
<p>The new audiences will automatically appear in the fullthrottle.ai Audience Library. The company&#8217;s AI Media Agent can also recommend them as marketers build campaigns.</p>
<p>That is significant for agencies and marketing teams that manage multiple campaigns. Audience creation can become another task on an already crowded media checklist. Having the platform surface relevant audiences can reduce some of that manual work.</p>
<p>There is also a practical benefit to keeping the audience inside the same platform used for campaign activation. Marketers can move from the original intent signal to additional media without having to recreate the audience across several systems.</p>
<h2>First-Party Data Is at the Center of the Strategy</h2>
<p>The announcement also fits into a larger shift in digital advertising toward first-party data.</p>
<p>First-party data is information a company collects directly from its own customer or prospect interactions. It can include website activity, customer records, purchase information, and other signals that an organization has a direct relationship with.</p>
<p>fullthrottle.ai positions its platform around first-party data and identity resolution, which is the process of connecting activity to a usable consumer or household identity.</p>
<p>The company says its platform does not rely on cookies and delivers audience match rates above 85 percent. It also uses SafeMatch attribution for measuring advertising outcomes.</p>
<p>Those capabilities are part of the company&#8217;s broader pitch to agencies, media companies, brands, and publishers: simplify the technology behind audience identification, activation, and measurement without forcing marketers to piece together several separate systems.</p>
<h2>Why the &#8220;Last Mile&#8221; Matters</h2>
<p>The name Last Mile Audiences is a useful description of what fullthrottle.ai is trying to solve.</p>
<p>The initial search or social interaction gets the consumer part of the way there. The consumer has shown interest. The advertiser has paid to generate that interest. But the person may not be ready to buy.</p>
<p>The next question is what the advertiser does with that signal.</p>
<p>Instead of treating the interaction as a completed transaction, Last Mile Audiences gives marketers another way to put that audience to work.</p>
<p>For a consumer who needs more time, repeated exposure across different channels may help keep the company in consideration. For an advertiser, that can mean getting more mileage from an investment that has already been made.</p>
<p>That is the part of the announcement that deserves attention. The feature is less about finding another place to run an advertisement and more about getting additional use from an audience that already demonstrated interest.</p>
<h2>fullthrottle.ai Is Targeting the Mid-Market</h2>
<p>fullthrottle.ai describes itself as an &#8220;AdTech Easy Button&#8221; for the mid-market.</p>
<p>The company combines AI-enabled identity resolution, campaign automation across multiple channels, and outcome-based measurement in one platform.</p>
<p>Its platform can turn website visitors into addressable households and activate first-party audiences across CTV, video, audio, display, and direct mail.</p>
<p>According to the company, more than 6,000 businesses nationwide use fullthrottle.ai.</p>
<p>For mid-market advertisers, that positioning is worth watching. Large enterprises can afford teams dedicated to audience management, media buying, analytics, identity resolution, and attribution. Smaller marketing organizations often have fewer people doing more jobs.</p>
<p>A platform that brings those functions together can make a difference if it reduces the amount of work required to move an audience from one advertising channel to another.</p>
<h2>Last Mile Audiences Is Available Now</h2>
<p>Last Mile Audiences is now available within the fullthrottle.ai platform.</p>
<p>The launch gives marketers a way to connect paid search and social engagement with broader media activation. The timing also makes sense as advertisers place greater value on first-party audiences, household-level targeting, cross-channel measurement, and the ability to use consumer signals beyond the platform where they were first collected.</p>
<p>The bigger lesson is straightforward. A search click is not necessarily a sale. A social engagement is not necessarily a conversion. But both can represent valuable intent.</p>
<p>fullthrottle.ai&#8217;s new approach is to keep that intent working after the original interaction. For marketers, that could turn a single click into a much longer advertising conversation.</p>
]]></content:encoded>
					</item>
	
			<item>
				<title>The Reputation Engagement I Would Rather Have Than the Friday Afternoon One</title>
				<link>https://www.billhartzer.com/marketing/unofficial-ballot-book/</link>
				<pubDate>Mon, 28 Sep 2026 18:26:01 +0000</pubDate>
								<dc:creator><![CDATA[Bill Hartzer]]></dc:creator>				<guid isPermaLink="false">https://www.billhartzer.com/?p=18531</guid>
					<description><![CDATA[A reporter emails your company at 4:40 on a Friday asking for comment on something...]]></description>

				<content:encoded><![CDATA[<p><img decoding="async" class="alignnone size-full wp-image-18532" src="https://www.billhartzer.com/wp-content/uploads/2026/09/the_unofficial_ballot-epub-cover.jpg" alt="The Unofficial Ballot book" width="1600" height="2560" srcset="https://www.billhartzer.com/wp-content/uploads/2026/09/the_unofficial_ballot-epub-cover.jpg 1600w, https://www.billhartzer.com/wp-content/uploads/2026/09/the_unofficial_ballot-epub-cover-188x300.jpg 188w, https://www.billhartzer.com/wp-content/uploads/2026/09/the_unofficial_ballot-epub-cover-640x1024.jpg 640w, https://www.billhartzer.com/wp-content/uploads/2026/09/the_unofficial_ballot-epub-cover-768x1229.jpg 768w, https://www.billhartzer.com/wp-content/uploads/2026/09/the_unofficial_ballot-epub-cover-960x1536.jpg 960w, https://www.billhartzer.com/wp-content/uploads/2026/09/the_unofficial_ballot-epub-cover-1280x2048.jpg 1280w" sizes="(max-width: 1600px) 100vw, 1600px" /></p>
<p>A reporter emails your company at 4:40 on a Friday asking for comment on something contested. You have about twelve hours before the absence of an answer becomes the answer. Most organizations spend those twelve hours arguing about what to say.<span id="more-18531"></span></p>
<p>The argument is the wrong one. It should have been settled months earlier.</p>
<p>That is the case I make in <em>The Unofficial Ballot: Trust, Risk, and Taking Sides in a Divided Era</em>, published by Hartzer Press. The Kindle edition went on sale September 24 and the paperback and hardcover followed on September 26. It runs 311 pages across six parts and 23 chapters, and it takes no position on any political question.</p>
<h2>Silence stopped being neutral</h2>
<p>For most of the internet era, saying nothing about a contested issue was read as having no position. That assumption is dead. Customers, employees, reporters and search engines now read silence as a statement, and they read it without asking what you intended.</p>
<p>The book is about one decision, made in advance: which issues your organization will speak to, which it will not, and on what basis. Write it down, apply it consistently, and you can explain it afterward to an audience with no interest in giving you the benefit of the doubt. Decide it in the moment and you are improvising in front of people taking screenshots.</p>
<h2>Why this landed on a search blog</h2>
<p>Because the record is searchable and it is now being summarized by machines.</p>
<p>Answer engines do not hand a user ten links to weigh. They name one brand, or they name none. SOCi&#8217;s 2026 analysis, which I covered in <a href="https://www.billhartzer.com/ai/ai-search-soci-top-100/">a look at the 100 brands AI search keeps choosing</a>, found systems increasingly returning a single recommendation rather than a ranked list. Visibility has turned into being picked.</p>
<p>Feed that same shift into a reputation question. When an AI system is asked what a company stands for, it composes an answer from whatever is public: the statement you made, the statement you did not make, the coverage of both, the employee posts, the review text. Nobody at the company approves that summary. Nobody sees it before a customer does.</p>
<p>An inconsistent public record used to produce an inconsistent impression. It now produces a confident paragraph, delivered as fact, generated from material you did not curate.</p>
<h3>Interpretation beats intent, every time</h3>
<p>A recurring theme of the book is the gap between what an organization meant and what its audience understood. Once a statement is public, only the second one operates. I have sat on calls with clients who wanted to litigate the first one. It is a comfortable conversation and it changes nothing.</p>
<p>I wrote about this after the Coldplay kiss cam episode, where <a href="https://www.billhartzer.com/reputation-management/coldplay-kisscam-chaos-andy-byron-online-reputation/">a few seconds of video became a permanent search result</a> for two executives and the company they worked for. Viral does not mean guilty. It does mean damage, and the damage runs on interpretation rather than on the facts anyone later establishes.</p>
<h3>Real backlash and manufactured outrage look identical for about a day</h3>
<p>This is the practical problem that catches communications teams, and the book gives it a chapter. Customer anger and a coordinated campaign produce the same dashboard in the first 24 hours. They call for opposite responses.</p>
<p>Respond to manufactured outrage as though it were customer sentiment and you hand it legitimacy along with a second news cycle. Dismiss real customer anger as astroturf and you have told the people who pay you that you were not listening. Telling them apart is a research question with a method, and getting it wrong is the most common expensive mistake I see.</p>
<h2>What I actually think about this</h2>
<p>Most crisis advice is written for companies large enough to have a communications department. That is a small slice of who gets hit.</p>
<p>The people who get hurt worst are small business owners whose reputation is local, personal and permanently indexed, and creators whose audience and livelihood are the same asset. Neither group has a playbook, and the standard advice — convene the team, align stakeholders, draft holding language — assumes resources they do not have. So the book builds the framework once and then runs it separately for four situations: executives and communications teams, small business owners, creators and public professionals, and anyone with a public account who has watched an old post age badly.</p>
<p>The other thing I believe, and it is less popular: the answer is often to say nothing, and that is a legitimate position when you reached it on purpose and can say why. The book is not an argument for speaking up. It is an argument against deciding by default.</p>
<h2>What to do before the next one</h2>
<p>You can do the first pass in an afternoon.</p>
<ul>
<li>Write down the three or four issues where your organization would be expected to have a view, and decide now, in writing, whether you speak to each one. Date the document.</li>
<li>Name who decides. Most bad statements come from a fast approval by whoever was reachable, not from a bad writer.</li>
<li>Search your own organization the way a customer does, including in an AI assistant, and read what comes back. That paragraph is your reputation to a growing share of people, and it is assembled from your public record whether you manage it or not.</li>
<li>Check what your own people have already said publicly under the company name. The statement you never made is rarely the exposure. The one somebody else made usually is.</li>
</ul>
<p>The book includes the worksheets and statement structures for all of this, and they are meant to be used rather than read. It is non-partisan by construction: written to be usable by readers who disagree with each other, because the process holds regardless of where anyone lands.</p>
<p>It is on Amazon in <a href="https://www.amazon.com/dp/B0HL2WG85N/">paperback</a>, <a href="https://www.amazon.com/dp/B0HL3KW1V5/">hardcover</a> and <a href="https://www.amazon.com/dp/B0HKWXBPW7/">Kindle</a>, with more about it at <a href="https://theunofficialballot.com/">theunofficialballot.com</a>. An audiobook edition follows.</p>
]]></content:encoded>
					</item>
	
			<item>
				<title>A Judge Ordered an Expert&#8217;s AI Prompts Produced. There Is Still No Standard, So I Wrote One</title>
				<link>https://www.billhartzer.com/ai/expert-record-book/</link>
				<pubDate>Mon, 28 Sep 2026 18:22:41 +0000</pubDate>
								<dc:creator><![CDATA[Bill Hartzer]]></dc:creator>				<guid isPermaLink="false">https://www.billhartzer.com/?p=18528</guid>
					<description><![CDATA[An expert witness sat for a deposition last year and was asked to produce the...]]></description>

				<content:encoded><![CDATA[<p><img decoding="async" class="alignnone size-full wp-image-18529" src="https://www.billhartzer.com/wp-content/uploads/2026/09/TER-Front-Cover-1800x2700-1-scaled.jpg" alt="The Expert Record book" width="1707" height="2560" srcset="https://www.billhartzer.com/wp-content/uploads/2026/09/TER-Front-Cover-1800x2700-1-scaled.jpg 1707w, https://www.billhartzer.com/wp-content/uploads/2026/09/TER-Front-Cover-1800x2700-1-200x300.jpg 200w, https://www.billhartzer.com/wp-content/uploads/2026/09/TER-Front-Cover-1800x2700-1-683x1024.jpg 683w, https://www.billhartzer.com/wp-content/uploads/2026/09/TER-Front-Cover-1800x2700-1-768x1152.jpg 768w, https://www.billhartzer.com/wp-content/uploads/2026/09/TER-Front-Cover-1800x2700-1-1024x1536.jpg 1024w, https://www.billhartzer.com/wp-content/uploads/2026/09/TER-Front-Cover-1800x2700-1-1365x2048.jpg 1365w" sizes="(max-width: 1707px) 100vw, 1707px" /></p>
<p>An expert witness sat for a deposition last year and was asked to produce the prompts he had typed into an AI system as he prepared his report. He did not have them. He had not kept them, because nobody had told him to, and no rule said he had to.<span id="more-18528"></span></p>
<p>That question is now ordinary. I have written a book about it.</p>
<p><em>The Expert Record: The Standard for Expert Witnesses Using AI</em> was published on September 27 by Hartzer Press, in paperback, hardcover and Kindle editions. It runs 274 pages across eight parts, 32 chapters and nine appendices, and it exists because the guidance available to an expert facing that question is either a ban nobody follows or an enthusiasm nobody can defend in front of a jury.</p>
<h2>What the book proposes</h2>
<p>At the center is the Expert Record Standard, version 1.0. Eight clauses, one page, written short enough to adopt and precise enough to test, so a violation is visible rather than a matter of opinion. The clauses cover sequence, non-delegation, independent authority, checkability before adoption, reproducible analysis, custody and confidentiality, contemporaneous record, and disclosure.</p>
<p>One governing test sits underneath all eight. An engagement satisfies the Standard when a competent examiner in the same field, working only from the primary sources and with no access to any AI system, could reproduce the expert&#8217;s result.</p>
<p>That test does most of the work. It does not ask whether a system was used. It asks whether the result depends on one.</p>
<p>Version 1.0 is licensed for adoption by individual experts and by firms, and the current text stays at <a href="https://theexpertrecord.com/standard/">theexpertrecord.com</a> rather than only in print, so a firm writing it into an engagement letter is copying from something maintained.</p>
<h2>Why this reaches past the courtroom</h2>
<p>I have been retained as an expert witness in matters involving SEO (Search Engine Optimization), paid search, web analytics, website traffic, domain names and internet investigations. Every one of those engagements now involves data productions too large to read page by page. That is where the pressure to use a system comes from, and it is not going away.</p>
<p>The same pressure reaches anyone whose signed work could become an exhibit. An agency that produces a traffic analysis for a client in a fee dispute. A consultant who writes a migration post-mortem that ends up attached to a complaint. A forensic accountant, an appraiser, a physician. The question is not whether you are a professional expert witness. It is whether your work product can be read aloud by someone trying to break it.</p>
<h3>Courts have started asking</h3>
<p>A magistrate judge in Connecticut ordered a party to revise its discovery responses to produce the AI prompts an expert&#8217;s team used to cull a document production, on the reasoning that an expert&#8217;s method is fair ground for discovery. That order was stayed on objection and had not been decided as of mid-September, so it is not settled law and I would not describe it as such. It is a signal, and it points one direction.</p>
<p>Trial courts have also split three ways on whether a person&#8217;s own AI session records are protected from discovery at all, and no United States court of appeals has answered the question. An expert planning around the most protective of those answers is planning around a coin flip.</p>
<h3>The failure mode is older than AI</h3>
<p>Expert testimony gets excluded for reasons that have nothing to do with technology and everything to do with what the opinion rests on. I wrote about the Federal Circuit ordering a new trial in <a href="https://www.billhartzer.com/google/new-trial-google-vs-ecofactor-thermostat/">Google v. EcoFactor</a>, where a damages expert built a per-unit royalty on contracts that did not say what he needed them to say, and leaned on what a CEO believed rather than what the documents showed.</p>
<p>No AI system was involved. The court&#8217;s objection was that the opinion could not be traced back to the record. Feed an unverified system output into the same report and you have the identical problem with a timestamp attached.</p>
<h2>The part people get wrong</h2>
<p>The industry keeps framing this as a technology question. It is not. The exposure is not the tool, it is the absence of a record.</p>
<p>An expert who can show what was asked, what came back, and what was done to check it has a method. An expert who cannot has an assertion. Every deposition I have watched turn on this has turned on that distinction, and the expert who lost it lost it in the first twenty minutes.</p>
<p>There is a second thing people get wrong, and it is more expensive. The instinct after a bad question is to shade the answer. Courts have been consistent about that in a way they have not been consistent about much else in this area: the tool draws a warning, and the answer about the tool draws the sanction. The Ninth Circuit suspended two lawyers for six months this year, and what moved the case from error to suspension was that they denied AI use when asked directly.</p>
<p>You do not get to decide, in the moment, how forthcoming to be. You decide it at retention, in writing, or it gets decided for you.</p>
<h2>What to do this week</h2>
<p>Three things, none of which require buying anything.</p>
<ul>
<li>Start a log today for every open engagement. Date, system, what it was used for, what you did to check the output. A spreadsheet is fine. The value is that it was made as the work proceeded, not reconstructed under an order.</li>
<li>Read the operative terms of service for the tier you actually use, and save a dated copy. Consumer and enterprise tiers differ on training, retention and confidentiality, and one federal court has already treated a provider&#8217;s published terms as the reason a user had no reasonable expectation of confidentiality.</li>
<li>Ask retaining counsel whether the protective order in your matter addresses AI tools. Some now bar open systems for every produced document, confidential or not. That question costs nothing before the work and a great deal after it.</li>
</ul>
<p>I wrote the book the same way I am asking you to work. I kept a contemporaneous log of my own AI use from before the first page of manuscript existed, and I published it in the appendix and on the companion site, because a book about contemporaneous records that did not keep one would be worth nothing.</p>
<p>The material that dates lives on the site rather than in the book: a record of decisions where expert AI use was actually at issue, the rules as they move, and my continuing log. The printed volume carries what will still be true when the current generation of systems has been replaced twice over.</p>
<p>The book is on Amazon in <a href="https://www.amazon.com/dp/B0HL5KZDHT/">paperback</a>, <a href="https://www.amazon.com/dp/B0HL5JG7H9/">hardcover</a> and <a href="https://www.amazon.com/dp/B0HL5RDH4D/">Kindle</a>. The Standard itself is free to read and free to adopt with attribution, which is the point of publishing it separately from the book.</p>
]]></content:encoded>
					</item>
	
			<item>
				<title>Facebook Says an Account Called “Meta Verrified” Sending Phishing Messages Doesn’t Violate Its Rules</title>
				<link>https://www.billhartzer.com/social-media/facebook-meta-verrified-account-phishing/</link>
				<pubDate>Fri, 18 Sep 2026 21:14:39 +0000</pubDate>
								<dc:creator><![CDATA[Bill Hartzer]]></dc:creator>				<guid isPermaLink="false">https://www.billhartzer.com/?p=18525</guid>
					<description><![CDATA[There are certain things you see on the internet that make you stop and wonder...]]></description>

				<content:encoded><![CDATA[<p><img loading="lazy" decoding="async" class="alignnone size-full wp-image-18524" src="https://www.billhartzer.com/wp-content/uploads/2026/09/Screenshot-2026-09-18-at-3.02.44-PM.png" alt="meta verrified account" width="935" height="710" srcset="https://www.billhartzer.com/wp-content/uploads/2026/09/Screenshot-2026-09-18-at-3.02.44-PM.png 935w, https://www.billhartzer.com/wp-content/uploads/2026/09/Screenshot-2026-09-18-at-3.02.44-PM-300x228.png 300w, https://www.billhartzer.com/wp-content/uploads/2026/09/Screenshot-2026-09-18-at-3.02.44-PM-768x583.png 768w" sizes="auto, (max-width: 935px) 100vw, 935px" /></p>
<p>There are certain things you see on the internet that make you stop and wonder if you&#8217;re missing something.</p>
<p><span id="more-18525"></span></p>
<p>This was one of them.</p>
<p>I recently came across a Facebook account using the name <strong>“Meta Verrified.”</strong> Yes, that&#8217;s “Verrified,” with two R&#8217;s.</p>
<p>The name immediately caught my attention because it looks remarkably similar to <strong>Meta Verified</strong>, Meta&#8217;s actual subscription and verification program.</p>
<p>But the name wasn&#8217;t the only issue.</p>
<p>The account was sending direct messages that appeared to be <strong>phishing attempts</strong>.</p>
<p>So I did what you would expect someone to do when encountering a suspicious account on Facebook.</p>
<p>I reported it to Meta.</p>
<p>And then I got Meta&#8217;s response.</p>
<h2>Facebook: “We didn&#8217;t find that Meta Verrified&#8217;s account went against our Community Standards”</h2>
<p>The response came through Facebook&#8217;s support system.</p>
<p>The message from Facebook says:</p>
<blockquote>
<p>“We didn&#8217;t find that Meta Verrified&#8217;s account went against our Community Standards.”</p>
</blockquote>
<p>It then says:</p>
<blockquote>
<p>“To keep our review process as fair as possible, we use the same set of Community Standards to review all reports.”</p>
</blockquote>
<p>Then comes the part that really caught my attention:</p>
<blockquote>
<p>“We&#8217;ve reviewed your report and found that the message doesn&#8217;t go against our Community Standards.”</p>
</blockquote>
<p>And that leaves me with a fairly obvious question:</p>
<p><strong>What exactly does have to happen before Facebook considers an account like this a problem?</strong></p>
<p>I&#8217;m not suggesting that Meta has determined that the account is legitimate. That&#8217;s not what the response says.</p>
<p>What it does say is that, after reviewing my report, Meta didn&#8217;t find the reported message to be a violation of its Community Standards.</p>
<p>That&#8217;s a pretty remarkable response considering the circumstances I reported.</p>
<h2>What makes “Meta Verrified” particularly interesting?</h2>
<p>Let&#8217;s start with the name.</p>
<p>The account isn&#8217;t called “John Smith&#8217;s Photography.”</p>
<p>It&#8217;s not called “Bob&#8217;s Facebook Page.”</p>
<p>It&#8217;s called <strong>“Meta Verrified.”</strong></p>
<p>The spelling is different from “Meta Verified,” but that&#8217;s arguably part of what makes names like this potentially confusing.</p>
<p>A person who sees the name quickly could conceivably interpret it as being associated with Meta&#8217;s verification program.</p>
<p>And that&#8217;s exactly the sort of thing that should cause people to slow down and look carefully at an account before interacting with it.</p>
<p>Meta itself acknowledges that scammers use impersonation and deceptive tactics on its platforms. In fact, Meta&#8217;s own safety materials specifically discuss scams in which people pretend to represent legitimate companies and send messages designed to get users to provide information or credentials.</p>
<p>Meta also says that fake accounts are often used by scammers and that it takes action against accounts associated with scam activity.</p>
<p>Meta&#8217;s published authenticity guidance also says that people on Facebook are expected to use the name they go by in everyday life and that pretending to be someone else or misrepresenting an authentic identity isn&#8217;t allowed.</p>
<p>That makes the response I received even more interesting.</p>
<h2>What is phishing?</h2>
<p>For anyone unfamiliar with the term, <strong>phishing is a form of online deception designed to trick someone into giving up sensitive information or taking an action that benefits the scammer.</strong></p>
<p>That information could include:</p>
<ul>
<li>A Facebook password</li>
<li>An email password</li>
<li>Credit card information</li>
<li>Banking information</li>
<li>Social Security numbers</li>
<li>Authentication codes</li>
<li>Other personal information</li>
</ul>
<p>A phishing message might pretend to come from a bank, technology company, government agency, friend or social media platform.</p>
<p>Meta specifically warns about <strong>user-support scams</strong>, where scammers pretend to work for companies such as Meta. According to Meta, these scammers may claim that there&#8217;s a problem with someone&#8217;s account and then send a phishing link designed to steal login credentials or even two-factor authentication codes.</p>
<p>That&#8217;s why a message that appears to come from an account associated with “Meta” deserves some scrutiny.</p>
<p>And it&#8217;s also why people should <strong>never assume that a message is legitimate simply because the account name looks official.</strong></p>
<h2>How to report a suspicious Facebook account</h2>
<p>If you encounter a Facebook account that appears to be fake, impersonating someone, phishing or attempting to scam people, there are several things you can do.</p>
<h3>1. Go directly to the account&#8217;s profile</h3>
<p>Don&#8217;t click on suspicious links sent to you by the account.</p>
<p>Instead, navigate directly to the Facebook profile in question.</p>
<p>Facebook provides reporting tools through its profile and content interfaces. Current reporting guidance also describes using the profile&#8217;s menu to select <strong>“Find support or report”</strong> and then selecting the appropriate reason for the report.</p>
<h3>2. Choose the reason that best describes the problem</h3>
<p>Depending on what you&#8217;re seeing, that might include an account that:</p>
<ul>
<li>Is pretending to be someone else</li>
<li>Is a fake account</li>
<li>Is using a fake name</li>
<li>Is attempting to scam people</li>
<li>Is sending suspicious messages</li>
<li>Is attempting to obtain personal information</li>
</ul>
<p>The important thing is to report the <strong>actual behavior</strong>, rather than simply reporting an account because you don&#8217;t like it.</p>
<h3>3. Report the individual message, too</h3>
<p>If the account sent you a suspicious Messenger message, don&#8217;t necessarily stop with the profile report.</p>
<p>Report the message itself.</p>
<p>That&#8217;s important because the suspicious behavior may be contained in the message rather than being obvious from the profile.</p>
<p>Meta says it provides reporting links on its platforms for abuse, harassment and other policy violations, and that Facebook users can monitor reports through the <strong>Support Inbox</strong>.</p>
<h3>4. Save evidence before reporting</h3>
<p>This is something I would strongly recommend.</p>
<p>Take screenshots.</p>
<p>Save the URL of the profile.</p>
<p>Save the messages.</p>
<p>If there&#8217;s a suspicious website involved, save the URL without visiting it if possible.</p>
<p>If the person is attempting to get you to enter a password, payment information or other sensitive information, document exactly what they&#8217;re asking for.</p>
<p>Why?</p>
<p>Because once an account or message disappears, you may no longer have access to the evidence.</p>
<h2>What if Facebook says the account doesn&#8217;t violate its rules?</h2>
<p>That&#8217;s essentially what happened to me.</p>
<p>Facebook&#8217;s own help documentation explains that after a report is reviewed, it will notify the person who made the report whether the reported content was determined to violate its Community Standards. Facebook also says that, in many cases, users can request another review if they disagree with the initial decision.</p>
<p>In my case, Facebook&#8217;s response included a <strong>“See options”</strong> button.</p>
<p>That is important.</p>
<p>If you report something and Facebook decides that it doesn&#8217;t violate its rules, check the available options rather than simply assuming the matter is closed.</p>
<p>According to Facebook&#8217;s current help information, the process is:</p>
<ol>
<li>Go to your Facebook profile menu.</li>
<li>Select <strong>Help &amp; support</strong>.</li>
<li>Select <strong>Support Inbox</strong>.</li>
<li>Find the report involving the other account.</li>
<li>Open Facebook&#8217;s decision.</li>
<li>Select <strong>See options</strong>.</li>
<li>If available, select <strong>Request another review</strong>.</li>
</ol>
<p>Not every report necessarily has the same options, so the choices displayed in your Support Inbox may vary.</p>
<h2>Don&#8217;t click the link just because the message looks official</h2>
<p>This may be the most important part of the whole thing.</p>
<p>If somebody sends you a Facebook or Messenger message saying that there&#8217;s a problem with your account, <strong>don&#8217;t click the link simply because the message appears to come from Meta.</strong></p>
<p>Go directly to Facebook yourself.</p>
<p>Open Facebook.</p>
<p>Check your notifications.</p>
<p>Check your account settings.</p>
<p>Check your Support Inbox.</p>
<p>In other words, verify the alleged problem independently instead of relying on the message to tell you how to fix it.</p>
<p>Meta itself recommends being suspicious of messages that ask you to provide passwords or other sensitive information and warns users not to click suspicious links or provide personal information in response to questionable requests.</p>
<h2>Meta is investing heavily in fighting scams. So what happened here?</h2>
<p>There&#8217;s another reason I found this particular response interesting.</p>
<p>Meta announced in March 2026 that it was expanding its anti-scam efforts across Facebook and Messenger, including additional scam detection, warnings and technology intended to detect impersonation and deceptive links. Meta said its systems analyze signals including text, images and surrounding context to help identify sophisticated scams.</p>
<p>That&#8217;s obviously a good thing.</p>
<p>But it also makes individual cases worth paying attention to.</p>
<p>Technology companies can have sophisticated automated detection systems and still have individual reports that don&#8217;t result in the outcome a user expects.</p>
<p>In this case, I reported an account called <strong>“Meta Verrified”</strong> that was sending messages I considered phishing attempts.</p>
<p>The answer I received was essentially:</p>
<p><strong>We looked at it. We don&#8217;t think it violates our rules.</strong></p>
<p>And that&#8217;s the part that I find difficult to understand.</p>
<p>I&#8217;m not claiming that Meta has endorsed the account.</p>
<p>I&#8217;m not claiming that the account is operated by Meta.</p>
<p>And I&#8217;m certainly not suggesting that a single report tells us how Meta&#8217;s entire enforcement system works.</p>
<p>But if an account is using a name that appears designed to resemble <strong>Meta Verified</strong> and is sending messages that appear to be phishing attempts, I think it&#8217;s entirely reasonable to ask why a report about that activity resulted in a finding that the message didn&#8217;t violate Meta&#8217;s Community Standards.</p>
<p>That&#8217;s exactly what makes this worth documenting.</p>
<p><strong>If Meta&#8217;s systems can&#8217;t identify something like this when a user specifically reports it, what is the average Facebook user supposed to do when they encounter a similar account?</strong></p>
<p>For now, the answer is pretty simple:</p>
<p><strong>Don&#8217;t trust the name. Don&#8217;t click the link. Don&#8217;t provide your credentials. Save the evidence. Report the account and the message. And if Facebook rejects the report, check whether you can request another review.</strong></p>
<p>Because sometimes the most important warning isn&#8217;t the suspicious message itself.</p>
<p>It&#8217;s the fact that someone had to report it in the first place.</p>
]]></content:encoded>
					</item>
	
			<item>
				<title>I Read 30 Digital Marketing Tool Licenses. Most Do Not Say What You Think.</title>
				<link>https://www.billhartzer.com/seo/digital-marketing-tool-licenses-what-you-can-publish/</link>
				<pubDate>Fri, 18 Sep 2026 00:26:16 +0000</pubDate>
								<dc:creator><![CDATA[Bill Hartzer]]></dc:creator>				<guid isPermaLink="false">https://www.billhartzer.com/?p=18515</guid>
					<description><![CDATA[You probably put a third-party traffic chart in a client deck this month. A Similarweb...]]></description>

				<content:encoded><![CDATA[<p><img loading="lazy" decoding="async" class="alignnone size-full wp-image-18520" src="https://www.billhartzer.com/wp-content/uploads/2026/09/hz_header_billhartzer.png" alt="" width="1600" height="840" srcset="https://www.billhartzer.com/wp-content/uploads/2026/09/hz_header_billhartzer.png 1600w, https://www.billhartzer.com/wp-content/uploads/2026/09/hz_header_billhartzer-300x158.png 300w, https://www.billhartzer.com/wp-content/uploads/2026/09/hz_header_billhartzer-1024x538.png 1024w, https://www.billhartzer.com/wp-content/uploads/2026/09/hz_header_billhartzer-768x403.png 768w, https://www.billhartzer.com/wp-content/uploads/2026/09/hz_header_billhartzer-1536x806.png 1536w" sizes="auto, (max-width: 1600px) 100vw, 1600px" /></p>
<p>You probably put a third-party traffic chart in a client deck this month. A Similarweb screenshot showing a competitor&#8217;s monthly visits, a Semrush visibility graph, an Ahrefs backlink count. I have done it hundreds of times.<span id="more-18515"></span></p>
<p>Here is a question I had never asked in 25 years of doing this work: does the license actually let me?</p>
<p>The question came up at BrightonSEO in San Diego a few days ago, in a conversation at a vendor booth about whether their data belongs in a legal case. I went back to my desk and read thirty license agreements covering the tools our industry runs on. Search visibility platforms, crawlers, link indexes, every major ad platform, analytics, panel measurement, domain forensics. All of them read in full, all of them coded the same way.</p>
<p>What I found matters more for the chart in your deck than for anything a court will ever see. I published the full findings as a <a href="/REPORT-LANDING-URL/">28-page research report</a>, and this is the part digital marketers need.</p>
<h2>Ten of Thirty Require Permission Before You Publish</h2>
<p>The clause that should worry you is not about lawyers. It is about presentation.</p>
<p>Section 6(iv) of the <a href="https://www.similarweb.com/corp/legal/terms/">Similarweb terms of service</a>, effective March 19, 2025, lists among the prohibited acts:</p>
<blockquote><p>present or share the data or information received through the Platform without Similarweb&#8217;s prior consent, and in the event such consent was given, present or share such data or information without attribution to Similarweb pursuant to Similarweb&#8217;s branding guidelines</p></blockquote>
<p>Read it as a marketer rather than as a litigator. A blog post is presenting data received through the Platform. So is a conference slide. So is a webinar, a pitch deck, a LinkedIn carousel, and a client proposal that gets forwarded to three people you have never met.</p>
<p>Ten of the thirty agreements I read require prior written consent to publish, or prohibit publication outright. SpyFu bars distribution without written consent, with a narrow exception that requires the phrase &#8220;Used with permission from SpyFu.com&#8221; on anything you do share. SparkToro prohibits reselling or white-labeling its output, then carves out an exception letting agencies use reasonable excerpts in client deliverables with attribution. Ahrefs bars publishing any portion of its services for commercial purposes, language written with resale in mind but broad enough that I would ask first.</p>
<p>Twelve of the thirty impose no publication restriction at all. Semrush is the most permissive of the visibility tools and bars sublicensing, scraping and AI training without restricting a chart. Your own Google Analytics and Search Console data carries no restriction whatsoever, because it is yours.</p>
<p>The practical fix takes one email. Vendors say yes, because a cited chart is free marketing. I have never had one refuse.</p>
<h2>Your Ad Performance Data May Not Be Yours</h2>
<p>This is the finding that surprised me most, and it affects every paid media team.</p>
<p>Amazon&#8217;s advertising agreement, posted July 23, 2026, states that all ad services data is Amazon&#8217;s exclusive property, restricts what you may do with it, and permits disclosure only where a valid and binding court order requires it, on condition that you tell Amazon before you disclose. LinkedIn&#8217;s ads agreement, updated November 3, 2025, expressly designates its metrics and professional demographics as confidential information and limits your use of ad services data to aggregate and anonymous performance assessment. TikTok&#8217;s commercial terms restrict sharing platform data with third parties.</p>
<p>Microsoft Advertising takes the opposite approach and includes no confidentiality clause at all. Meta does not designate campaign data as confidential, and publishes political advertisers&#8217; spend, reach and targeting in its Ad Library for seven years.</p>
<p>So two of the four largest ad platforms in the world claim your campaign performance data and two do not. Nobody mentions which is which when you open the account. If you publish client benchmarks, run a public performance study, or put spend figures in a case study, that distinction decides whether you are inside the agreement.</p>
<h2>The Accuracy Gap Nobody Talks About</h2>
<p>Twenty-seven of the thirty agreements disclaim the accuracy of their data. Three describe that data as estimates or modeled output anywhere in their published terms.</p>
<p>Those are not the same thing, and the difference is the whole argument. A warranty disclaimer is a liability instrument. It says the vendor will not be answerable if you rely on the number and lose money. A statement of measurement uncertainty is a methodological instrument. It tells you what the number is, how it was produced, and how far it is likely to sit from the thing it estimates.</p>
<p>Similarweb states that its data rests on estimations and extrapolations. Google publishes a caveat that Trends data should be treated as one data point among others and is not a perfect mirror of search activity. Nielsen publishes real sampling error documentation. The other twenty-seven give you a disclaimer and nothing else. Majestic comes closest to candor without using the word, explaining that the shifting structure of the web makes it &#8220;impossible to provide completely accurate link data.&#8221;</p>
<h2>How Far Off Are the Estimates? There Is an Actual Answer</h2>
<p>Our industry argues about this in blog comments. A peer-reviewed study settled it with data.</p>
<p>Jansen, Jung and Salminen, publishing in <a href="https://doi.org/10.1371/journal.pone.0268212">PLOS ONE in 2022</a>, compared Similarweb against the sites&#8217; own Google Analytics across 86 websites in 26 countries and 19 industry verticals over twelve months. Similarweb understated total visits by 19.4 percent and unique visitors by 38.7 percent. It overstated bounce rate by 25.2 percent and average session duration by 56.2 percent.</p>
<p>The number that matters is not the size of the gap. It is the shape. The bias runs in a consistent direction, which means these estimates are not noisy, they are skewed, and a skew you can name is a skew you can account for. Traffic reads low. Engagement reads high.</p>
<p>The problem is not unique to one vendor. Research at the ACM Internet Measurement Conference in 2022 checked popularity rankings against Cloudflare server-side request data and found 87.1 percent of the Alexa top thousand were overranked, with 56.7 percent overranked by two or more orders of magnitude.</p>
<p>None of that makes the tools useless. It makes them what they have always been: modeled approximations that are excellent for direction and pattern and unfit to carry a precise factual claim on their own. Say &#8220;Similarweb estimates roughly 2 million visits&#8221; and you are describing the instrument correctly. Say &#8220;they get 2.1 million visits a month&#8221; and you have made a statement the vendor declined to make.</p>
<h2>Where This Gets Expensive</h2>
<p>The legal side is my day job, so one paragraph on it. A terms of service is a contract between you and a vendor. It cannot make a document inadmissible, and it is not a defense to a subpoena. That is governed by the rules of evidence and decided by a judge. What a restrictive license can do is put you in breach, which costs you the account or a damages claim, and it can collide with the disclosure obligations that apply when you are retained as an <a href="https://www.billhartzer.com/seo-expert-witness/">expert witness</a>. Exactly one of the thirty agreements addresses litigation at all, and that one prohibits it.</p>
<p>The report covers all of that in detail. For most readers here, the courtroom is not the risk. The risk is publishing something your license did not permit, or publishing an estimate as though it were a measurement, in front of an audience that includes your next client.</p>
<h2>What I Would Do Monday Morning</h2>
<p>Read the licenses for the tools you actually publish from. It took me an afternoon for thirty of them, and I had been paying for some for a decade without opening the agreement once.</p>
<p>Ask permission where the terms require it, and keep the reply. Follow the attribution format the vendor specifies, because several of them dictate the wording.</p>
<p>Label estimates as estimates, with the tool, the metric, the date range and the known limitation in the same paragraph as the figure. A footnote is not a disclosure. When I <a href="https://www.billhartzer.com/seo/analyzed-20-years-backlinks-half-gone/">analyzed twenty years of my own backlinks</a>, I published the index, the export, the statistical method and the limitation that biases the result, and the limitation was the part readers wrote to me about. Disclosure did not weaken the analysis. It was the reason anyone trusted it.</p>
<p>Apply the same skepticism to other people&#8217;s numbers. I <a href="https://www.billhartzer.com/ai/semrush-study-chatgpt-topics/">took apart a Semrush study of ChatGPT categories</a> not because the data was bad but because the headline number everyone repeated described something other than what readers assumed.</p>
<p>And build the important work on data you own. Your analytics, your Search Console, your ad account exports, your server logs, your own crawl. Nobody licenses those to you and nobody can revoke them. That is true whether you are diagnosing a <a href="https://www.billhartzer.com/website-traffic-rankings-loss-audit/">traffic and rankings loss</a> or writing the study that wins you a speaking slot.</p>
<p>The tools are good. The claims we make about them are often worse than the tools deserve. Every platform in this survey measures something real by a method with known limits, and those limits are not a defect. They are the specification.</p>
<p><strong>The full research report is free.</strong> Thirty license agreements, the coding for each, the evidence rules, the reliability research and the complete source list. <a href="https://www.hartzer.com/digital-marketing-tool-licenses-study/">Download the 28-page PDF here</a>.</p>
<p><script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "NewsArticle",
  "mainEntityOfPage": {
    "@type": "WebPage",
    "@id": "https://www.billhartzer.com/seo/digital-marketing-tool-licenses-what-you-can-publish/"
  },
  "headline": "I Read 30 Digital Marketing Tool Licenses. Most Do Not Say What You Think.",
  "description": "I read 30 digital marketing tool licenses. Ten need permission before you publish their data. 27 disclaim accuracy. Only 3 call it an estimate.",
  "image": {
    "@type": "ImageObject",
    "url": "https://www.billhartzer.com/wp-content/uploads/2022/07/cropped-bhlogo-1200x1200-1.jpg",
    "width": 1200,
    "height": 1200
  },
  "author": {
    "@type": "Person",
    "name": "Bill Hartzer",
    "url": "https://www.billhartzer.com/about/",
    "jobTitle": "Search Engine Marketing and Domain Name Consultant",
    "worksFor": {
      "@type": "Organization",
      "name": "Hartzer Consulting"
    },
    "sameAs": [
      "https://www.linkedin.com/in/billhartzer/",
      "https://x.com/bhartzer"
    ]
  },
  "publisher": {
    "@type": "Organization",
    "name": "Bill Hartzer",
    "url": "https://www.billhartzer.com/",
    "logo": {
      "@type": "ImageObject",
      "url": "https://www.billhartzer.com/wp-content/uploads/2022/07/cropped-bhlogo-1200x1200-1.jpg",
      "width": 1200,
      "height": 1200
    }
  },
  "datePublished": "2026-09-17T09:00:00-05:00",
  "dateModified": "2026-09-17T09:00:00-05:00",
  "articleSection": "SEO",
  "keywords": "terms of service, SEO tools, digital marketing data, Similarweb, Semrush, web analytics, data licensing, competitive analysis, ad platform data",
  "about": [
    {
      "@type": "Thing",
      "name": "Digital marketing tools"
    },
    {
      "@type": "Thing",
      "name": "Terms of service"
    },
    {
      "@type": "Thing",
      "name": "Web analytics"
    }
  ],
  "citation": "https://doi.org/10.1371/journal.pone.0268212"
}
</script></p>
]]></content:encoded>
					</item>
	
			<item>
				<title>131 SEO Experts Just Ranked Google&#8217;s Ranking Factors. Backlinks Did Not Die.</title>
				<link>https://www.billhartzer.com/seo/2026-google-ranking-factors-survey/</link>
				<pubDate>Wed, 09 Sep 2026 18:02:49 +0000</pubDate>
								<dc:creator><![CDATA[Bill Hartzer]]></dc:creator>				<guid isPermaLink="false">https://www.billhartzer.com/?p=18510</guid>
					<description><![CDATA[Cyrus and Dawn Shepard published the 2026 Google Ranking Factors Expert Survey this morning. One...]]></description>

				<content:encoded><![CDATA[<p><img loading="lazy" decoding="async" src="https://www.billhartzer.com/wp-content/uploads/2026/09/linkedin-square-1200x1200-1.png" alt="" width="2400" height="2400" class="alignnone size-full wp-image-18511" srcset="https://www.billhartzer.com/wp-content/uploads/2026/09/linkedin-square-1200x1200-1.png 2400w, https://www.billhartzer.com/wp-content/uploads/2026/09/linkedin-square-1200x1200-1-300x300.png 300w, https://www.billhartzer.com/wp-content/uploads/2026/09/linkedin-square-1200x1200-1-1024x1024.png 1024w, https://www.billhartzer.com/wp-content/uploads/2026/09/linkedin-square-1200x1200-1-150x150.png 150w, https://www.billhartzer.com/wp-content/uploads/2026/09/linkedin-square-1200x1200-1-768x768.png 768w, https://www.billhartzer.com/wp-content/uploads/2026/09/linkedin-square-1200x1200-1-1536x1536.png 1536w, https://www.billhartzer.com/wp-content/uploads/2026/09/linkedin-square-1200x1200-1-2048x2048.png 2048w" sizes="auto, (max-width: 2400px) 100vw, 2400px" /></p>
<p>Cyrus and Dawn Shepard published the <a href="https://signal.zyppy.com/p/google-ranking-factors-expert-survey">2026 Google Ranking Factors Expert Survey</a> this morning. One hundred thirty-one SEO professionals scored more than 100 possible ranking factors on a seven-point scale, from strongly positive to strongly negative, producing 13,665 data points and about 43 hours of collective work.</p>
<p><span id="more-18510"></span></p>
<p>The top of the list will disappoint anyone hoping for a revolution. Relevance came first, named in the top three by 57.1 percent of respondents. Backlinks came second at 54.8 percent. Content quality was third at 47.6 percent.</p>
<p>Same three as a decade ago. Which is the story, though not the one most coverage will run with.</p>
<h2>What 131 Practitioners Put At The Top</h2>
<p>The full top ten, by the share of experts naming each factor among their three most important:</p>
<ul>
<li>Relevance, 57.1 percent</li>
<li>Backlinks, 54.8 percent</li>
<li>Content quality, 47.6 percent</li>
<li>Authority and trust, 36.5 percent</li>
<li>Behavior and click signals, 29.4 percent</li>
<li>Brand signals, 27.0 percent</li>
<li>User satisfaction, 19.8 percent</li>
<li>Technical SEO health, 17.5 percent</li>
<li>Topical authority, 14.3 percent</li>
<li>Internal links, 11.1 percent</li>
</ul>
<p>Melissa Popp put the relevance finding in the plainest terms available: &#8220;You can have technically perfect content with pristine headings and a gorgeous URL structure, but if the person searching wanted one thing and you answered another, none of that saves you.&#8221; Every SEO has watched a beautifully built page lose to an ugly one that answered the question. That is the whole finding.</p>
<p>Meta description scored lowest among the relevance factors. Twenty years of clients asking me to rewrite them, and the people who do this for a living rate them near zero for rankings. They still earn clicks, which is a separate argument worth having.</p>
<h2>The Two Findings That Actually Moved</h2>
<h3>Behavior signals came out into the open</h3>
<p>Click and behavior signals landed fifth, and user satisfaction seventh. Ten years ago saying that out loud got you an argument. AJ Kohn&#8217;s contribution was four words long: &#8220;Navboost is still a monster.&#8221;</p>
<p>What changed is not Google. What changed is the documentation. The antitrust testimony and the leaked Content Warehouse API material gave the industry names and mechanisms for behavior it had suspected for years, and the survey reflects a profession that has stopped hedging. Kohn&#8217;s practical advice follows from it: build your own satisfaction metric, page type by page type, and treat it as a ranking input rather than a reporting nicety.</p>
<p>Gerry White drew the useful line. Bounce rate is a bad metric, he said, but users who pogo-stick back to the results page and pick a competitor are sending a signal Google takes seriously. Return to SERP was in fact the lowest-rated user signal in the survey, scored as an active negative.</p>
<p>Diego Ivo asked the question the whole industry is going to be arguing about by next year: when most browsing is done by agents rather than people, how does anyone measure user satisfaction at all?</p>
<h3>Links now do double duty</h3>
<p>Backlinks held second place, which by itself is just continuity. The interesting part is what one is now worth.</p>
<blockquote><p>&#8220;With AI, links can have an even greater effect. A client recently got mentioned and linked from a Reuters article a few weeks ago. Their visibility in ChatGPT went up 4x over the next week,&#8221; said Andrew Shotland.</p></blockquote>
<p>That is a single practitioner reporting a single case, so hold it loosely. It also matches what everyone working on AI visibility has been seeing. A link from a trusted publication now feeds two systems at once: Google&#8217;s ranking of your page and the corpus that language models pull from when someone asks for a recommendation. The same placement, twice the return.</p>
<p>On what makes a link count, Nitin Manchanda was blunt: &#8220;A hundred links from real audiences beat a thousand links nobody ever clicks.&#8221; Links from trusted domains and topically relevant pages scored highest. Spam links scored as strong negatives. Geoff Kenyon added the anchor text nuance that trips people up, which is that exact match anchors help to a point and suppress performance past it, and the variable that matters is what share of your profile they represent rather than the raw count.</p>
<h2>The Domain Name Finding Nobody Will Lead With</h2>
<p>Buried in the relevance section is a line from Shai Aharony: &#8220;I&#8217;m just as surprised as everyone else by how effective EMDs are still.&#8221;</p>
<p>EMD stands for exact match domain, a domain name that is the keyword itself. Google released an algorithm update targeting low-quality EMDs in 2012, and the industry has been declaring them dead ever since. Fourteen years later, working SEOs are still surprised to find them working.</p>
<p>I am not surprised, and I have <a href="https://www.billhartzer.com/domain-names/new-gtld-ranking/">documented exact match domains ranking on new gTLDs</a>, where designertrunks.shop sits at the top for its keyword. What the 2012 update actually targeted was thin sites hiding behind a keyword domain. It never removed the benefit of a name that plainly says what the business is. A relevant, memorable domain earns better anchor text, better branded search, better click-through from the results page, and more trust from anyone deciding whether to link. Those are four of the factors on the list above.</p>
<p>Domain choice is not a ranking factor in the way a title tag is. It is an input to half a dozen of them, which is more useful and much harder to undo later.</p>
<h2>Where The Experts Disagreed</h2>
<p>The disagreements are the most valuable part of a survey like this, because consensus findings tend to be the ones you already knew.</p>
<p>Site speed and Core Web Vitals produced the sharpest split. Ryan Jones called them &#8220;the most over-rated things in SEO,&#8221; arguing the page either loads or it does not. Greg Gifford answered with context: speed matters enormously for a large ecommerce site, and a local car dealership with heavy vehicle photography operates under different expectations. Both are right, which is exactly why blanket ranking factor advice fails.</p>
<p>AI-generated content split the room in a more interesting way. The survey found it viewed negatively only where it was scaled with little added value. Abby Gleason reported good results from high-quality AI content at scale on a site with substantial unique material and heavy human editing. Ross Simmonds framed the reason it cuts both ways: &#8220;AI has made content cheap. Anyone can spin up a thousand articles overnight. And Google knows it. That&#8217;s why quality is becoming the filter.&#8221; Gianluca Fiorelli named the trap on the other side, that too much consensus makes you invisible and too much contrarianism gets you buried.</p>
<p>Technical SEO landed eighth, which the report characterizes as table stakes. Marshall Simmonds said it best: &#8220;Tech SEO is a commodity at this point. It won&#8217;t necessarily win the race, but it can absolutely lose it.&#8221; Dana DiTomaso supplied the proof, describing a client whose canonical tags pointed at unrelated pages and who saw roughly a 60 percent lift in impressions and clicks once it was fixed. That is not a ranking factor doing work. That is <a href="https://www.billhartzer.com/seo/tech-seo-conflicting-signals/">a self-inflicted wound being closed</a>, and I find one on most sites I audit.</p>
<h2>What This Survey Can And Cannot Tell You</h2>
<p>Here is the part to keep in mind before anyone builds a strategy deck out of it.</p>
<p>This is a survey of expert opinion. It is not a correlation study, and it is not controlled testing. It measures what 131 experienced practitioners believe, and belief moves for reasons that have nothing to do with the algorithm changing. Behavior signals did not necessarily get more important this year. What happened is that a court case and a document leak made it professionally safe to say they were important all along.</p>
<p>That distinction matters when you are deciding where to spend. The survey is excellent evidence of where informed practitioners are placing their bets, and Cyrus has never claimed more than that. Treat it as the profession&#8217;s collective read rather than a specification, and it earns its place.</p>
<p>The consistency is what I take from it. Relevance, links, quality and trust have topped every credible version of this exercise going back to the Moz surveys, through Panda, Penguin, mobile-first, Core Web Vitals, helpful content and now AI answers. Jan-Willem Bobbink went furthest on brand, arguing a site can survive being technically broken and poorly optimized as long as people search for and talk about it. Roger Montti compressed the same idea into a sentence: branded search is the new link.</p>
<p>Neither man was describing a tactic. They were describing what happens when a business is worth talking about, which is the one ranking factor that has never appeared on any list and has decided the outcome every time.</p>
<p>Read the survey. Then go fix your canonicals.</p>
<p><script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "NewsArticle",
  "mainEntityOfPage": {
    "@type": "WebPage",
    "@id": "https://www.billhartzer.com/seo/2026-google-ranking-factors-survey/"
  },
  "headline": "131 SEO Experts Just Ranked Google's Ranking Factors. Backlinks Did Not Die.",
  "description": "Zyppy's 2026 Google Ranking Factors Survey polled 131 SEOs across 13,665 data points. Relevance and links still lead, and exact match domains still work.",
  "image": {
    "@type": "ImageObject",
    "url": "https://www.billhartzer.com/wp-content/uploads/2022/07/cropped-bhlogo-1200x1200-1.jpg",
    "width": 1200,
    "height": 1200
  },
  "author": {
    "@type": "Person",
    "name": "Bill Hartzer",
    "url": "https://www.billhartzer.com/about/",
    "jobTitle": "Search Engine Marketing and Domain Name Consultant",
    "worksFor": {
      "@type": "Organization",
      "name": "Hartzer Consulting"
    },
    "sameAs": [
      "https://www.linkedin.com/in/billhartzer/",
      "https://x.com/bhartzer"
    ]
  },
  "publisher": {
    "@type": "Organization",
    "name": "Bill Hartzer",
    "url": "https://www.billhartzer.com/",
    "logo": {
      "@type": "ImageObject",
      "url": "https://www.billhartzer.com/wp-content/uploads/2022/07/cropped-bhlogo-1200x1200-1.jpg",
      "width": 1200,
      "height": 1200
    }
  },
  "datePublished": "2026-09-09T15:00:00-05:00",
  "dateModified": "2026-09-09T15:00:00-05:00",
  "articleSection": "SEO",
  "keywords": "Google ranking factors, SEO survey, Zyppy, Cyrus Shepard, backlinks, Navboost, exact match domains, click signals",
  "about": [
    {
      "@type": "Thing",
      "name": "Google ranking factors"
    },
    {
      "@type": "Person",
      "name": "Cyrus Shepard"
    },
    {
      "@type": "Organization",
      "name": "Zyppy",
      "url": "https://zyppy.com/"
    }
  ],
  "citation": "https://signal.zyppy.com/p/google-ranking-factors-expert-survey"
}
</script></p>
]]></content:encoded>
					</item>
	
			<item>
				<title>AI Local Business Recommendations Went From 6% To 45% In A Year. The Bigger Number Is 88%</title>
				<link>https://www.billhartzer.com/local-search/ai-local-business-recommendations-45-percent/</link>
				<pubDate>Wed, 09 Sep 2026 17:44:50 +0000</pubDate>
								<dc:creator><![CDATA[Bill Hartzer]]></dc:creator>				<guid isPermaLink="false">https://www.billhartzer.com/?p=18508</guid>
					<description><![CDATA[Forty-five percent of U.S. consumers used an AI tool to get a local business recommendation...]]></description>

				<content:encoded><![CDATA[<p><img loading="lazy" decoding="async" src="https://www.billhartzer.com/wp-content/uploads/2026/09/ai-search-pdf-preview-image.webp" alt="AI search study" width="500" height="707" class="alignnone size-full wp-image-18505" srcset="https://www.billhartzer.com/wp-content/uploads/2026/09/ai-search-pdf-preview-image.webp 500w, https://www.billhartzer.com/wp-content/uploads/2026/09/ai-search-pdf-preview-image-212x300.webp 212w" sizes="auto, (max-width: 500px) 100vw, 500px" /></p>
<p>Forty-five percent of U.S. consumers used an AI tool to get a local business recommendation in the past twelve months. Last year that figure was 6 percent.</p>
<p><span id="more-18508"></span></p>
<p>That is the number worth your attention this week, and it did not come from the press release announcing it. It comes from BrightLocal&#8217;s <a href="https://www.brightlocal.com/research/lcrs-ai-trust/">Local Consumer Review Survey 2026</a>, which found that AI has become the third most common source of local business recommendations, behind Google and Facebook and ahead of everything else. Cybertegic, a Pasadena digital marketing agency, built a white paper around those findings and issued a release this morning pointing to them.</p>
<p>A seven-fold jump in one year is the steepest adoption curve local search has produced in a long time. So take the finding seriously. Then read the rest of the study, because the part nobody is quoting changes what you should do about it.</p>
<h2>What The BrightLocal Data Actually Says</h2>
<p>The survey ran through a SurveyMonkey panel of 1,002 U.S. adults. The AI questions went to the 455 respondents who said they had used AI for local recommendations, and BrightLocal published that cut on March 10, 2026, on a separate page from the main survey the release cites.</p>
<p>Among those 455 people:</p>
<ul>
<li>63 percent said they trust the recommendations AI gives them</li>
<li>71 percent said they trust AI-generated review summaries</li>
<li>88 percent said they check the sources or verify the legitimacy of what AI tells them</li>
<li>97 percent said they at least sometimes cross-check an AI recommendation against real reviews</li>
<li>42 percent said they always verify on the review platform itself</li>
</ul>
<p>On which tools they reach for, ChatGPT leads at 31 percent of consumers, Google AI Mode follows at 23 percent, and Gemini, Copilot and Claude trail behind both.</p>
<h2>Why 88 Percent Is The Number That Should Change Your Plan</h2>
<p>Read two of those findings next to each other. Sixty-three percent trust AI recommendations. Eighty-eight percent go check them anyway.</p>
<p>That is not a contradiction. It is a description of how people actually behave, and it demolishes the framing this story usually gets. AI is not replacing the way consumers evaluate local businesses. It is becoming the front door that sends them to the same places they were already going.</p>
<p>Somebody asks ChatGPT for a personal injury attorney in Fort Worth. It names three firms. Then, per this data, roughly nine out of ten of those people open a new tab and look one of those firms up. What do they find? Your Google Business Profile. Your reviews. Your site. The same assets that decided the outcome before any of this started.</p>
<p>So the practical takeaway runs opposite to the panic. Your review corpus and your <a href="https://www.billhartzer.com/local-search/removed-google-business-profile-again-why-matters/">Google Business Profile</a> did not become less load-bearing when AI entered local search. They became the verification layer for a channel that now sends 45 percent of consumers your way. Being named by ChatGPT and then failing the review check is a worse outcome than never being named, because you paid for the impression and lost the customer at the last step.</p>
<h2>What The Guide Says AI Weighs</h2>
<p>Before it gets to strategies, the white paper lists six signals it says conversational systems evaluate when picking what to reference:</p>
<ul>
<li><strong>Authority.</strong> Does the brand show real expertise and trustworthiness?</li>
<li><strong>Clarity.</strong> Is the information direct and well organized?</li>
<li><strong>External validation.</strong> Do reputable websites reference the brand consistently?</li>
<li><strong>Structured data.</strong> Can machines interpret the content without guessing?</li>
<li><strong>Topical depth.</strong> Does the site cover the subject thoroughly?</li>
<li><strong>Brand consistency.</strong> Do your business signals line up across the web?</li>
</ul>
<p>Now set that beside the guide&#8217;s list of why businesses fail to show up in AI answers: thin service pages, weak entity signals, inconsistent business information, no schema, minimal digital PR, and blog posts published at random to chase keywords. The two lists describe the same website from opposite ends. One is what a good audit looks for. The other is what it finds.</p>
<h2>The Eight Strategies, Named</h2>
<p><a href="https://www.cybertegic.com/resources/white-paper/ai-visibility-optimization-guide/">Cybertegic&#8217;s AI Search Survival Guide</a> sits behind an email form, and its landing page promises eight strategies without listing them. Here they are, with what the guide actually asks you to do under each:</p>
<ul>
<li><strong>Strengthen entity-based SEO.</strong> Keep the business name identical everywhere, optimize the Google Business Profile, and make every listing match exactly, so systems read one business rather than three similar ones.</li>
<li><strong>Build consistent multi-platform brand signals.</strong> Reviews, directory listings, social profiles, press mentions, community discussion and third-party citations, all describing you the same way.</li>
<li><strong>Build topical authority.</strong> Pick three to five core topic areas, write one pillar page per major service, publish eight to twelve supporting articles under each, and interlink them.</li>
<li><strong>Publish answer-focused content.</strong> Turn FAQs into standalone articles, use question-based headings, and answer in the opening line before adding the detail.</li>
<li><strong>Structure content for machine readability.</strong> Headings every 200 to 300 words, paragraphs of two to four sentences, the direct answer at the top of each section, comparison tables where they earn their space.</li>
<li><strong>Implement schema markup.</strong> LocalBusiness, Organization, FAQ, Article and Product types, matching what visitors can see, validated on a schedule instead of set once.</li>
<li><strong>Earn mentions across trusted websites.</strong> Digital PR, guest articles, podcast appearances and industry directories, counting brand mentions as authority even where no link follows.</li>
<li><strong>Optimize for generative engines.</strong> Write passages that can be quoted whole, favor plain definitions over keyword stuffing, and format for extraction.</li>
</ul>
<p>Now count how many of those you have not heard for a decade. I get none. The pillar-and-supporting-articles prescription is a content silo with a new label. Consistent citations, schema, digital PR and question-led headings have been on every competent local SEO checklist for years.</p>
<p>To the guide&#8217;s credit, it says so itself. Its GEO section closes by stating that GEO is not replacing SEO. That is the honest position, and it sits oddly next to a document sold on the premise that AI visibility is a distinct practice. Google has said the same thing, and <a href="https://www.billhartzer.com/seo/google-confirmed-aiseo-seo/">so have I</a>: the generative features run on core Search ranking and quality systems, so the existing best practices carry over. When someone quotes this guide at you and a separate retainer in the same breath, that closing line is the sentence to read back to them.</p>
<p>The genuinely useful idea in the white paper is not a strategy at all. It is the reframe of what you measure: keywords become entities, rankings become citations, backlinks become multi-source authority, and click-through traffic becomes inclusion in the answer itself. Your business can be the source behind a recommendation and never receive the click.</p>
<p>That is not hypothetical. The guide cites SparkToro&#8217;s finding that nearly 60 percent of Google searches already end without anyone clicking through, and AI answers push in the same direction. A local business can be gaining influence over buying decisions during a quarter when its analytics show traffic sliding, and nothing in a standard monthly report will explain the gap. Of the five trends the guide forecasts for the next two to three years, that one is worth taking seriously: brands competing for citations rather than positions. The rest read like safe extrapolation.</p>
<h2>Where The Release Gets Ahead Of Itself</h2>
<p>The case studies prove less than they appear to. The release reports that a probate mortgage financing client showed up in a Google AI Overview for a Pasadena search, and a family law firm turned up in ChatGPT recommendations for a Monterey Park query. I believe both happened. I also cannot check either one, and neither can you. AI answers vary by account, location, session, phrasing and hour, and the release concedes as much in its own disclaimer. A screenshot of one AI answer is an anecdote, not a ranking, and our industry is about to spend two years pretending otherwise.</p>
<p>The Gartner forecast has run out of runway. The guide warns that Gartner predicted a 25 percent decline in traditional search volume by 2026. It is September 2026. A prediction whose deadline has arrived stops being a forecast and becomes a testable claim, and a 2026 white paper citing it as a future risk without checking the scoreboard is doing marketing, not analysis.</p>
<p>None of which makes the guide bad. It leans on Semrush, HubSpot, Ahrefs and McKinsey rather than original research, which is normal for the genre, and its 90-day plan is sensibly ordered: audit your entity consistency, schema and service page depth first, build content clusters second, chase mentions third. An agency publishing that and asking for your email is an honest trade. Just separate the survey data, which is evidence, from the packaging, which is a sales argument.</p>
<h2>What I Would Do First</h2>
<p>Ask the machines about yourself. Open ChatGPT and Google AI Mode, ask for the best businesses in your category in your city, and run it five or six times with different phrasing. Write down who gets named and which sources get cited. Nearly half of American consumers have run a query like that in the past year. Most business owners have never once seen the answer.</p>
<p>Then check what AI says about your facts, not your ranking. Hours, service area, whether you take a particular insurance, what you charge. AI systems state these things with total confidence and get them wrong often enough that businesses have been left arguing with customers over offers they never made. Wrong information about you is a more urgent problem than absent information about you.</p>
<p>After that, work the guide&#8217;s month one and stop. Audit your name, address and phone across every platform that describes you. Check that your schema matches your visible pages. Find the service you sell most and see whether its page says more than a paragraph and a form. That is a week of work for most businesses, it feeds AI answers and traditional rankings at the same time, and it costs nothing beyond attention.</p>
<p>Skip the separate AI visibility retainer until someone shows you a measurement method that survives being run twice.</p>
<p>Local search has absorbed shifts like this before. Maps, mobile, the local pack, zero-click results. Each arrived with a wave of consultants selling a new discipline, and each time the winners were the businesses with accurate information, real reviews and pages that answered the question.</p>
<p>Forty-five percent of consumers are now asking a chatbot where to go. Almost all of them are checking its homework before they spend a dollar. Make sure that when they check, they find you looking exactly like the business the machine said you were.</p>
<p><script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "NewsArticle",
  "mainEntityOfPage": {
    "@type": "WebPage",
    "@id": "https://www.billhartzer.com/local-search/ai-local-business-recommendations-45-percent/"
  },
  "headline": "AI Local Business Recommendations Went From 6% To 45% In A Year. The Bigger Number Is 88%.",
  "description": "BrightLocal found 45% of US consumers used AI for local business recommendations, up from 6%. But 88% verify what AI tells them. That changes the playbook.",
  "image": {
    "@type": "ImageObject",
    "url": "https://www.billhartzer.com/wp-content/uploads/2022/07/cropped-bhlogo-1200x1200-1.jpg",
    "width": 1200,
    "height": 1200
  },
  "author": {
    "@type": "Person",
    "name": "Bill Hartzer",
    "url": "https://www.billhartzer.com/about/",
    "jobTitle": "Search Engine Marketing and Domain Name Consultant",
    "worksFor": {
      "@type": "Organization",
      "name": "Hartzer Consulting"
    },
    "sameAs": [
      "https://www.linkedin.com/in/billhartzer/",
      "https://x.com/bhartzer"
    ]
  },
  "publisher": {
    "@type": "Organization",
    "name": "Bill Hartzer",
    "url": "https://www.billhartzer.com/",
    "logo": {
      "@type": "ImageObject",
      "url": "https://www.billhartzer.com/wp-content/uploads/2022/07/cropped-bhlogo-1200x1200-1.jpg",
      "width": 1200,
      "height": 1200
    }
  },
  "datePublished": "2026-09-09T13:00:00-05:00",
  "dateModified": "2026-09-09T13:00:00-05:00",
  "articleSection": "SEO",
  "keywords": "local SEO, AI search, BrightLocal, Local Consumer Review Survey, Google Business Profile, ChatGPT local recommendations, Google AI Overviews, Cybertegic",
  "about": [
    {
      "@type": "Thing",
      "name": "AI-assisted local business discovery"
    },
    {
      "@type": "Organization",
      "name": "BrightLocal",
      "url": "https://www.brightlocal.com/"
    },
    {
      "@type": "Organization",
      "name": "Cybertegic",
      "url": "https://www.cybertegic.com/"
    }
  ],
  "citation": "https://www.brightlocal.com/research/lcrs-ai-trust/"
}
</script></p>
]]></content:encoded>
					</item>
	
			<item>
				<title>IAB: 56% Want Creator Voices In AI Shopping Advice, But Nobody Can Verify Which Creators Are Honest</title>
				<link>https://www.billhartzer.com/advertising/iab-creator-signal-ai-shopping-recommendations/</link>
				<pubDate>Wed, 09 Sep 2026 17:23:30 +0000</pubDate>
								<dc:creator><![CDATA[Bill Hartzer]]></dc:creator>				<guid isPermaLink="false">https://www.billhartzer.com/?p=18506</guid>
					<description><![CDATA[The Interactive Advertising Bureau (IAB) released research this morning showing that 56 percent of consumers...]]></description>

				<content:encoded><![CDATA[<p><script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "NewsArticle",
  "mainEntityOfPage": {
    "@type": "WebPage",
    "@id": "https://www.billhartzer.com/advertising/iab-creator-signal-ai-shopping-recommendations/"
  },
  "headline": "IAB: 56% Want Creator Voices In AI Shopping Advice, But Nobody Can Verify Which Creators Are Honest",
  "description": "IAB research finds 56% of consumers want creator perspectives in AI shopping recommendations. The trust signals they name are the ones AI cannot verify.",
  "image": {
    "@type": "ImageObject",
    "url": "https://www.billhartzer.com/wp-content/uploads/2022/07/cropped-bhlogo-1200x1200-1.jpg",
    "width": 1200,
    "height": 1200
  },
  "author": {
    "@type": "Person",
    "name": "Bill Hartzer",
    "url": "https://www.billhartzer.com/about/",
    "jobTitle": "Search Engine Marketing and Domain Name Consultant",
    "worksFor": {
      "@type": "Organization",
      "name": "Hartzer Consulting"
    },
    "sameAs": [
      "https://www.linkedin.com/in/billhartzer/",
      "https://x.com/bhartzer"
    ]
  },
  "publisher": {
    "@type": "Organization",
    "name": "Bill Hartzer",
    "url": "https://www.billhartzer.com/",
    "logo": {
      "@type": "ImageObject",
      "url": "https://www.billhartzer.com/wp-content/uploads/2022/07/cropped-bhlogo-1200x1200-1.jpg",
      "width": 1200,
      "height": 1200
    }
  },
  "datePublished": "2026-09-09T09:00:00-05:00",
  "dateModified": "2026-09-09T09:00:00-05:00",
  "articleSection": "Digital Marketing",
  "keywords": "IAB, AI shopping recommendations, creator marketing, influencer marketing, AI search visibility, online reviews, CreatorFronts",
  "about": [
    {
      "@type": "Organization",
      "name": "Interactive Advertising Bureau",
      "url": "https://www.iab.com/",
      "sameAs": "https://en.wikipedia.org/wiki/Interactive_Advertising_Bureau"
    },
    {
      "@type": "Thing",
      "name": "AI-assisted shopping recommendations"
    }
  ],
  "citation": "https://www.iab.com/insights/the-creator-signal/"
}
</script></p>
<p><img loading="lazy" decoding="async" src="https://www.billhartzer.com/wp-content/uploads/2024/03/iab_Logo-300x150.jpg" alt="iab logo" width="300" height="150" class="alignnone size-medium wp-image-13696" srcset="https://www.billhartzer.com/wp-content/uploads/2024/03/iab_Logo-300x150.jpg 300w, https://www.billhartzer.com/wp-content/uploads/2024/03/iab_Logo.jpg 640w" sizes="auto, (max-width: 300px) 100vw, 300px" /></p>
<p>The Interactive Advertising Bureau (IAB) released research this morning showing that 56 percent of consumers prefer AI-generated shopping recommendations that include creator or influencer perspectives. Sixty-five percent say they feel more confident in an AI recommendation when credible creator reviews sit behind it.</p>
<p><span id="more-18506"></span></p>
<p>The study, called <a href="https://www.iab.com/insights/the-creator-signal/">The Creator Signal</a>, surveyed 2,200 people in the United States, United Kingdom, Australia, Mexico and India, all screened for having used AI for shopping research in the previous three months. IAB published it six days before its first <a href="https://www.iab.com/events/creatorfronts-2026/">CreatorFronts</a>, an invite-only marketplace in New York where platforms, creator marketplaces and measurement vendors pitch media buyers.</p>
<p>Keep that second fact in view. It shapes how the first one should be read.</p>
<h2>What The Research Found</h2>
<p>The appetite splits hard by age. About 65 percent of Gen Z and Millennials want creator perspectives folded into AI recommendations, against 53 percent of Gen X and 34 percent of Boomers. Among Gen Z, a creator review inside an AI answer carries almost the same weight as brand or advertiser information, 27 percent against 29 percent.</p>
<p>Eighty percent of respondents said they are comfortable relying on AI recommendations at all. That is the number that should stop you before any of the creator findings do.</p>
<p>On which creators people trust, fame finished last:</p>
<ul>
<li>Everyday consumers, at 52 percent</li>
<li>Professional reviewers, at 43 percent</li>
<li>Subject-matter experts, at 37 percent</li>
</ul>
<p>On what makes a creator trustworthy enough for an AI system to lean on:</p>
<ul>
<li>Credibility, at 46 percent</li>
<li>Consistency, at 40 percent</li>
<li>Expertise, at 36 percent</li>
<li>Independence from brands and clear sponsorship disclosure, at 34 percent</li>
</ul>
<p>Shoppers see the most value in creator-informed AI when they are hunting for the best price (38 percent), finding products they did not know existed (32 percent), comparing options (31 percent) and boiling down product information (31 percent). India and Mexico showed the strongest appetite of the five markets. The United States led the Western ones. The United Kingdom was the most cautious.</p>
<blockquote><p>&#8220;What we&#8217;re seeing is that with AI shopping, consumers don&#8217;t just want the recommendation. They want real-world experience and expertise behind the guidance,&#8221; said Jack Koch, SVP of Research and Insights at IAB.</p></blockquote>
<h2>Why This Matters If You Sell Anything Online</h2>
<p>For twenty-five years the job was to make your own pages rank, because the page was the answer. That arrangement is coming apart in front of us.</p>
<p>Ask an AI assistant which cordless drill to buy and the model is not reading your product page back to you. It is assembling a verdict out of everything it has ingested about cordless drills: YouTube teardowns, Reddit threads, review-site roundups, forum complaints, comparison posts written by people you have never heard of and cannot call. Your marketing copy is one voice in that pile, and it is the voice with the weakest claim on the model&#8217;s trust.</p>
<h3>Your off-site reputation is now your product copy</h3>
<p>This is the shift most brand teams have not budgeted for. You can control your site. You cannot control the twelve-minute video where somebody takes your product apart on a kitchen table, and that video may carry more weight in an AI answer than your entire domain. <a href="https://www.billhartzer.com/ai/viral-nation-ai-discovery-brands/">Visibility inside AI answers</a> now depends on what other people publish about you, in places you do not own, on a schedule you do not set.</p>
<p>That should feel familiar. It is link building&#8217;s logic applied to sentiment instead of anchor text, and it rewards what it always rewarded: being worth talking about.</p>
<h3>Sponsored and independent look identical to a machine</h3>
<p>Thirty-four percent of respondents said independence from brands and honest sponsorship disclosure is what makes a creator worth trusting. Fair enough. Now name the system that can tell the difference at scale.</p>
<p>A paid review and an unpaid one are the same words in the same order. The FTC disclosure lives in a spoken aside at 0:14 or in the fourth line of a video description, and both routinely vanish when the content gets transcribed, scraped, summarized or syndicated. By the time that text reaches a training corpus or a retrieval index, the sponsorship marker is usually gone.</p>
<h2>Where The Study Stops Short</h2>
<p>I have watched survey findings get sold as market facts for a long time, and this one needs three qualifiers the press release does not supply.</p>
<p>Stated preference is not behavior. People say they want independent expertise the same way they say they read terms of service. What they actually do is buy the first thing the assistant names.</p>
<p>The most-trusted category is the most gamed one. Everyday consumers finished first at 52 percent, and everyday consumer reviews are the single most polluted content type on the internet. The Transparency Company flagged 14 percent of the 73 million reviews it analyzed as fake or highly suspicious, and put the consumer harm from <a href="https://www.billhartzer.com/marketing-foo/300-billion-fake-reviews/">review fraud</a> at $300 billion a year across home services, legal and medical alone. Three percent of those reviews were written by AI, a share that has been growing roughly 80 percent month over month since mid-2023. The people running those operations have already worked out that the audience worth fooling is no longer human. Feed an AI system &#8220;authentic everyday consumers&#8221; as a trust signal and you have handed them a bigger payday than they ever had in search.</p>
<p>The 72 percent figure needs a footnote. IAB reports that nearly three-quarters of consumers use general-purpose AI assistants for shopping decisions, but the sample here was screened for people who had already used AI for shopping research. The release does not say whether that number comes from this survey or a separate general-population study. The gap between those two readings is enormous, and any marketer quoting the figure in a budget meeting should find out which one it is first.</p>
<p>None of that makes the research wrong. David Cohen, IAB&#8217;s CEO, said trust will be a critical differentiator, and he is right. It makes the research a trade association&#8217;s opening argument, published a week before the trade association sells influencer inventory to media buyers in Midtown. Read it as evidence, not as a verdict.</p>
<h2>What To Do About It</h2>
<p>Start by finding out what the assistants already say about you. Ask ChatGPT, Gemini, Copilot and Perplexity to recommend a product in your category, then ask them to compare your product against your two closest competitors. Write down which sources they name. That list is your new backlink profile, and most brands have never looked at it.</p>
<p>Then go work the sources. Get your products into the hands of reviewers who publish where the models actually read: YouTube, established review sites, active subreddits, category forums with years of archived threads. Make sponsorship disclosure loud and early in the content itself rather than buried in a description field, because clean disclosure is what survives being scraped.</p>
<p>Keep your structured product data correct. Schema markup for price, availability, ratings and specifications is one of the few inputs you fully control, and retrieval systems lean on it when they need a fact rather than an opinion.</p>
<p>The brands that win the next stretch of this will be the ones with the most credible third-party evidence attached to their names, not the ones with the best on-page copy. That is a harder job. It is also a more durable one, because you cannot buy it in a quarter.</p>
<p>IAB&#8217;s five markets all agreed on the same underlying point, and it is not really a finding about AI. Shoppers want to know that a person handled the thing before they spend money on it. The machine in the middle is new. What people want from it is very old.</p>
]]></content:encoded>
					</item>
	
			<item>
				<title>Semrush Says Showing Up Once In A ChatGPT Category Is Worse Than Not Showing Up</title>
				<link>https://www.billhartzer.com/ai/semrush-study-chatgpt-topical-focus/</link>
				<pubDate>Thu, 03 Sep 2026 20:47:21 +0000</pubDate>
								<dc:creator><![CDATA[Bill Hartzer]]></dc:creator>				<guid isPermaLink="false">https://www.billhartzer.com/?p=18499</guid>
					<description><![CDATA[Semrush released a second study of ChatGPT topic behavior on August 3, and it answers...]]></description>

				<content:encoded><![CDATA[<p><img loading="lazy" decoding="async" class="alignnone size-full wp-image-18500" src="https://www.billhartzer.com/wp-content/uploads/2026/09/semrush-topical-focus-square-1080.png" alt="SEMrush topical focus ai study" width="1080" height="1080" srcset="https://www.billhartzer.com/wp-content/uploads/2026/09/semrush-topical-focus-square-1080.png 1080w, https://www.billhartzer.com/wp-content/uploads/2026/09/semrush-topical-focus-square-1080-300x300.png 300w, https://www.billhartzer.com/wp-content/uploads/2026/09/semrush-topical-focus-square-1080-1024x1024.png 1024w, https://www.billhartzer.com/wp-content/uploads/2026/09/semrush-topical-focus-square-1080-150x150.png 150w, https://www.billhartzer.com/wp-content/uploads/2026/09/semrush-topical-focus-square-1080-768x768.png 768w" sizes="auto, (max-width: 1080px) 100vw, 1080px" /></p>
<p>Semrush released a second study of ChatGPT topic behavior on August 3, and it answers the question the first one left hanging. The July research told us that domain-level SEO metrics do a poor job of predicting who owns a topic. This one identifies what does a better job, and the answer has an uncomfortable edge to it.<span id="more-18499"></span></p>
<p>The headline finding for anyone planning a content calendar: a brand that appeared in only one of a category&#8217;s five prompts was associated with a <em>drop</em> in mention share. Not a modest gain. A drop. That penalty stayed in place until the brand reached at least three of the five prompts.</p>
<p>Partial coverage of a category looks worse than staying out of it.</p>
<p>The full <a href="https://www.semrush.com/blog/chatgpt-topical-focus-study/?utm_source=influencers&amp;utm_medium=BillHartzer&amp;utm_campaign=topics-first-research2">Semrush study on how topical authority spreads in ChatGPT</a> is worth reading alongside my write-up of <a href="https://www.billhartzer.com/ai/semrush-study-chatgpt-topics/">the earlier topic ownership research</a>, because the two fit together.</p>
<h2>What The New Study Measured</h2>
<p>Margarita Loktionova and Kevin Indig of Growth Memo worked from the same tracking base as before. The same 1,094 US categories, five prompt variants each, monitored monthly from January through June 2026. This round produced 283,215 domain-category citation observations and 76,493 brand-mention observations, plus 45,578 category-expansion appearances mapped to 1,458 brand entities.</p>
<p>Two definitions carry the analysis. A brand&#8217;s core expertise was any category where it had already shown up in at least three of the five prompts before the month being measured. Category closeness was scored by semantic similarity between the target category&#8217;s prompts and the categories the brand already answered well.</p>
<p>That second definition is the part most write-ups will skip past, and it is the mechanism behind everything that follows.</p>
<h2>Citations Travel. Brand Names Stay Home.</h2>
<p>In categories close to a brand&#8217;s core expertise, brands were cited as a source in 74% of appearances, named in the answer text in 44%, and both cited and named in 34%.</p>
<p>Move out to categories distant from that core and the numbers collapse to 50% cited, 25% named, and 9% both.</p>
<p>The conversion rate tells the same story more sharply. Among citations in the least related categories, 18% also carried a named brand. In the most relevant categories, 46% did. Roughly two and a half times the rate, and the both-cited-and-mentioned gap of 34% against 9% is close to a fourfold spread.</p>
<p>Indig put the practical version plainly:</p>
<blockquote><p>Being cited in many categories does not show a spread-thin effect in this sample. But brand recognition works differently. The categories where a brand consistently earns named mentions are the ones from which it can credibly expand.</p></blockquote>
<h2>Why The Two Signals Behave So Differently</h2>
<p>Here is my read on the mechanism, and it is the piece I have not seen anyone else state.</p>
<p>Citations and mentions come out of two different systems. A citation is a retrieval outcome. A document matched the query, cleared the source bar for relevance, authority and freshness, and got linked. Retrieval scores documents. It has no opinion about whether your company belongs in a category. Publish a competent article about a subject you have no business in and retrieval will happily surface it.</p>
<p>A mention is a generation outcome. The model names a brand because its internal representation associates that entity with that domain of expertise. Entities, not pages.</p>
<p>That distinction explains the whole distance curve. Documents port across categories because document scoring is local to the document. Entity associations do not port, because they were built from everything the model has absorbed about who you are. It also explains the finding from the July study that only 21% of most-cited domains were also the most-mentioned brand, with a slightly negative correlation of -0.229. That number always read as strange to me. This study shows it was an average taken across a distribution with real structure inside it. Close to your core, the two signals converge. Far from it, they come apart.</p>
<h2>Legal And Healthcare Break The Pattern</h2>
<p>Industry changed the shape of the results. In finance and real estate, broader category coverage came with more citation leverage. Legal and healthcare behaved differently. Citation gains from wider coverage were weaker, and even at full five-prompt coverage, mention share still trended negative.</p>
<p>Anyone who has worked on medical or legal sites through the last decade of Google quality updates will recognize that fingerprint. These are the categories where search engines have applied a raised bar for years, and the model appears to carry a similar reluctance to name a brand in high-stakes regulated subjects no matter how thoroughly that brand covers the questions.</p>
<p>I want to flag that as inference rather than finding. The study did not test author credentials, licensure signals or primary research, and the authors said so. If you sell into legal or healthcare, coverage depth reads as table stakes in this data, not as the lever. The lever is probably credentialing and third-party validation, and somebody should test that properly.</p>
<h2>The Question I Would Ask Before Acting On Any Of This</h2>
<p>Every relationship in this study is observational. No controls, no intervention, and the authors chose the phrase &#8220;associated with&#8221; carefully.</p>
<p>So here is the question that keeps me from taking the depth finding at face value. Is shallow coverage causing the drop in mention share, or is shallow coverage a symptom of being a small brand? A company that appears in one of five prompts in a category is plausibly a company nobody has heard of. It would show a weak mention share whether or not it published anything else. Reverse causality is very much alive in that chart, and the one-of-five penalty may be measuring brand size wearing a coverage costume.</p>
<p>The 1,458 mapped brand entities are worth noting too. The July study drew on more than 50,000 brands. The expansion analysis here rests on a much narrower base, and narrow bases produce confident-looking curves.</p>
<p>None of that makes the work less valuable. It makes it a set of hypotheses good enough to plan against and not yet good enough to bet the quarter on.</p>
<h2>What This Changes About Category Planning</h2>
<p>The old playbook was to publish one piece into a new category, watch what happens, and decide whether to commit. This data says that test costs you something and teaches you very little, because a single appearance sits below the threshold where the signal turns positive.</p>
<p>Treat three of five prompts as your minimum viable entry. Below that, you are buying a result the study associates with going backward.</p>
<p>Pick your next category by proximity rather than by traffic estimate. Semantic similarity to what you already answer well is the variable that moved the numbers here, and you can approximate it yourself. Embed your existing prompt sets, embed the candidate category&#8217;s prompts, rank the candidates by cosine similarity, and work outward from the top. That is a Tuesday afternoon of Python, not a platform purchase.</p>
<p>Then feed the mention side of the equation, because that is the side that gates expansion. Clear positioning, consistent entity signals, and coverage from the sources the model actually reads: review platforms, industry publications, communities, expert commentary.</p>
<p>Topical authority buys you a running start into the category next door. It does not buy you a category three doors down, and the data says the attempt may cost you ground in the process.</p>
]]></content:encoded>
					</item>
		</channel>
</rss>
<!-- last GN Pub feeds fetch (not specifically this feed): 2026-10-31 01:45:06 -->
