For our own web sites and our clients’ sites, we monitor new links and links lost. It’s a good practice for any SEO. More importantly, we monitor when Google recognizes or disregards incoming links to a website as reported in Google Webmastertools.

We’ve noticed these events seem to occur in clusters, and last night indicated that Google is thinking less of links from Twitter. In one sweep, Google dropped the Twitter Profile “URL” incoming link for every site we monitor.

Examples from a few of our own sites:

Granted, the links from Twitter pages are nofollow. However the majority of sites that are active on Twitter see a subsequent boost in rankings on Google (we’ve seen the same). We’ll continue to monitor whether or not Google re-recognizes these links in the future.

UPDATE: July 1, 2009

Over the last three weeks, we have seen a gradual re-inclusion of the links from Twitter. At this point all original links are reinstated. Interesting. Very interesting.

a

Google seems to be dropping links from Twitter

It’s very simple to view this new data:

Select the “Search Query Report” under the “Reports” tab in your AdWords account.

In the example above, we are prompted to place a negative match on the keyword “Free”.

Use this report on phrase match and broad match groups to find negative keywords as well as surprise phrases that generate better-than-average conversion.

a

AdWords Now Shows Actual Phrase Used for Click

Anyway, here I am shopping for a new chain. Riders become very attuned to the feel of the shifting, and getting exactly the right chain that shifts smoothly is a matter of considerable experimentation and rest-stop chatter. I’m Googling for a KMC chain - prized for the special coating of 11 herbs and spices, I assume.

I click an AdWords link on the SERPs ‘KMC chain shimano 10spd’ and discover the retailer is breaking a fundamental rule of PPC ads: Show the products the consumer just searched for. Instead, I’m dumped into a page where I re-select the brand I want. I click KMC and find they don’t have the 10 speed version in stock - only the 9 speed. As I reach for the back button I momentarily switch from bike geek to adwords geek, and think to myself ‘yikes, I’ll bet their bounce rate numbers are scary’. Only I then realize they’re not.

The catch is that bounce rate in Google Analytics means ’saw only one page, then left’. But I saw two pages because I was forced to select the brand before seeing the products. It’s a bad campaign, leading to a bad landing page and then a product that cannot be purchased, but bounce rate won’t reveal that. Ouch.

a

Dreadful Landing Pages Lead to Low Bounce Rate. HUH?

On the other hand, those doing SEO will repeatedly hear that you need good copy in your pages so Googlebot knows what the page is talking about.

So there lies the paradox: Googlebot is the only entity that can be bothered to read your carefully crafted copy, and even it does it solely to work out whether your page is relevant for a searcher who has a vanishingly small attention span.

a

People Don’t Read Copy; Only Googlebot Has Time

Google then exploited these links as a way of determining which page is most relevant, creating the famous PageRank mechanism.

I’m wondering if people use these in-content hyperlinks less these days. After all, relevant stuff is just a search away. With the advent of browser toolbar search boxes, it’s even easier for people to search instead of using hyperlinks.

Will content writers continue to link to stuff, as the use of those links decline?

In many ways Twitter adds to the problem for Google. Lots of links point to a page but there’s no anchor text and the content is very transient.

This might start to undermine a fundamental aspect of determining relevance, eroding the quality of the results. Or are they already so washed-out that it makes no difference?

a

Is the hyperlink headed for extinction?

Nice move Google! Segmentation rocks!

In my testing, patience ran out at 45 statements which tested as accurate.

I used a 10,297 character regular expression* in one of the statements and it tested out just fine.

Have you gone higher than that?

* I’ll share my simple tool to translate a list into a regular expression for Google Analytics.

a

Limits on Google Analytics Segmentation

Here’s a “way back” trip to my former dream team.

Bryan of ZURB interaction designers just posted some images that remind me of the privilege I had to work with a highly creative and fun team at ClickTracks - back in the day.

Blue guy represented a web visitor in an analytics world, but grew to represent reports, products and promotions over time.

We went a little overboard trying to come up with a blue guy to represent the “Dissection Report”. You could imagine some of the prototypes. No need. There’s enough here to get the idea across. Thanks Bryan.

a

Memories of a former dream team

The script measures the likelihood that an email will fall into a recipient’s “junk mail” folder in Outlook 2003. Outlook 2003 uses a word weighting system and this measurement does not guarantee delivery into other mail clients. However, the spam words indicated by Outlook 2003 are used in other inbox protection systems and we have found this tool to be very useful to help us clean up copy and increase the chance for delivery into all inboxes.

Paste your email subject and text in below. A low (negative) spam score means the email is likely to be received. A positive score means the email is less likely to make it into the inbox. Replace as many red flashing words as possible, add neutral and ‘good’ words until your spam score is less than zero

This idea and original code came from our friend Tim Cadell who can be thanked at t cadell at gmail. It’s unmodified other than a total score display.

a

Outlook 2003 Spam Predictor

http://www.slideshare.net/johnbuckman/employees-suck-presentation?type=powerpoint

a

How to bootstrap a company