1. Quality score affects your ad position.

OK, most people know this, but it’s worth mentioning just in case: Google determines the position of your ad based on a keyword’s ad rank, not based on your bid. Ad rank is calculated as quality score multiplied by your bid.

If your quality score for a keyword is high as compared to your competition’s quality score for that keyword, your ad may appear above theirs even if you’re bidding less than they are.

2. The biggest factor in determining Quality Score is your CTR for the keyword.

The higher your CTR, the higher your Quality Score.

3. CTR is relevant to the specific search network where your ad is shown.

When your ad shows on a Google search site (e.g. google.com, google.ca) only your CTR on Google search sites will affect the quality score used to compute ad rank.

When your ad shows on a search partner site (e.g. ask.com or aol.com) your CTR on that specific partner site and your CTR across the search network affect the quality score used to compute ad rank.

Display network CTR doesn’t affect the quality score used for the search network.

4. CTRs are normalized for ad position.

All things being equal, if your ad is shown in a higher position, it will have a higher CTR. Google obviously knows this and therefore considers your CTR relative to your ad position when determining a keyword’s quality score. From a quality score standpoint, you aren’t penalized for being in a lower ad position.

• Google compares your CTR when shown in position 1 to how competitors perform for that keyword in position 1.
• Google compares your CTR when shown in position 2 to how competitors perform for that keyword in position 2.
• And so on.

5.  Landing page quality is either relevant or not relevant.

When determining the quality score of a keyword, its landing page is judged as either relevant or not relevant.  If the landing page is considered relevant (“no problems”), there is nothing more you can do to the landing page that would further optimize the keyword’s quality score.

6.  Landing page load time is either good or poor.

Similar to the above, once your landing page load time is deemed fast enough to be good (“no problems”), speeding up the load time even more won’t help the keyword’s quality score.

7. Quality score is based on exact match.

If you have broad, phrase, and exact match versions of a keyword, only the exact match CTR affects quality score. If you have only phrase and/or broad match versions, then only the CTR for search queries that match the keyword exactly are used.

A note about Quality Score history

Only a keyword’s current quality score is shown in the AdWords interface.  To keep track of quality score changes, each week run (and save) a keyword report that includes the quality score for each of your keywords.

Suggested Reading:

• More on Ad Rank [adwords.google.com]

I was recently befuddled when I embarked on some keyword research and found the Google External Keyword Tool was showing only 11 months of data. My client was adamant that I take trends for the prior month into consideration.

Without prior month data in the Google Keyword Tool, I felt like Popeye without spinach. Now what? As I started muttering curses to Google under my breath, it occurred to me to check Google Insights.

Google Insights for Search allows you to compare search volume patterns “across specific regions, categories, time frames and properties.”  In my case, I was interested in seeing how search patterns changed over time. Data in Insights is up-to-date, not just to the prior month, but up to the current day.

The caveat is Insights doesn’t provide absolute numbers. The data is normalized and displayed on a scale of 1 – 100, thereby giving you relative trend comparisons, rather than absolute search volumes or absolute comparisons.

Absolute data would be nice, but my foremost interest was in how search terms were trending based on recent news coverage. Plus, by combining the somewhat outdated absolute search volumes I received from the Google External Keyword Tool with the up-to-date relative trends from Google Insights, I could nail down some reasonable recommendations for the client.

Above, is a graph I generated from Google Insights.  I cropped out the legend, which decodes what each of the colored lines represents, to protect the anonymity of my client.  But that decoding wouldn’t add much here anyway.

Each line in the graph represents a different search term.  You can see how relative search volumes for those terms have changed over the past 12 months.

• A few of the terms peaked in March
• The blue term spiked again in April
• In late December, the blue and orange terms had another spike

Based on this, I was able to inform the client which terms were heating up again in Google searches.  To get an idea of not only the relative volume, but the actual volume for December, I looked at absolute search volumes (from the External Keyword Tool) for earlier months and scaled relative December data accordingly.

There is one simplification in the description above compared to my actual method. Prior to comparing terms with Insights I grouped all keywords under consideration (roughly 200) into 8 themes.  I then chose a representative keyword for each theme by identifying the keyword within that theme with the highest average absolute search volume for the past year (according to the External Keyword Tool).  So in the graph above, I was actually comparing themes based on the representative keywords.

Some other nice Google Insights features, which I didn’t use here, but could come in handy:

• You’ll notice a “News Headlines” checkbox above the graph in my screenshot above.  Click this and Google will correlate various points of time in the graph with news stories.
• A “Regional Interest” section appears below the graph and compares interest levels by country or sub-region.  I did a test comparing “chevy volt” and “nissan leaf” searches for the last 90 days and this feature revealed Michigan, Wisconsin, and Indiana to be hot for the Chevy Volt, but Hawaii, Washington and Oregon hot for the Nissan Leaf.
• A “Rising Searches” section tells you which searches have experienced growth in the given time period, as compared to the preceding time period. For example, “chevy volt price” (+140%), “nissan leaf mileage” (+100%) and “nissan leaf mpg” (+70%) were the top 3 for the chevy volt / nissan leaf comparison.

Give it a whirl yourself: Google Insights for Search

In my previous post, I discussed how the Google External Keyword Tool was showing only 11 months of data.  Finally today (January 27) data for the 12 month (December) started appearing.

Why did it take so long? I have no idea. Starting now, I intend to take note each month of when the prior month’s search volume comes online and will share that information on this blog.

UPDATE: Today (2/18/2011) is the first day I’ve seen January 2011 data available. Prior month data appeared 9 days faster than last month.

UPDATE 2: (3/2/2011) I checked to see if February data is available yet.  It isn’t.  But here’s something interesting: the Local Search Trends column is gone. What’s up with that?

UPDATE 3: (3/17/2011) First day that February data is available. Luck of the Irish.

UPDATE 4: (4/1/2011) As noted in the comments section of this post, all month-to-month data stopped being available sometime last week.  There is a thread on Google’s forums about this, including the following explanation:

Hi All,

In an effort to provide the best experience for our Keyword Tool users, we often experiment with ways to offer statistics that are simpler to look at and interpret.

When we noticed that most advertisers weren’t using the “Local Search Trends” column in the Keyword Tool, we decided to experiment with removing the column.

If you do use the local search trends data, I’d ask that you please fill out the survey linked from the Keyword Tool (link: https://spreadsheets3.google.com/a/google.com/viewform?formkey=dHAzTkJ2dm1sZnlWbnhpOVpXNURPbXc6MA#gid=0), and describe how you used the local search trend data. We use the survey to find ways to improve the tool and get a better idea of the type of statistics that are helpful from a wider range of users.

I’d also like the open the thread up to a discussion of how people on the forum generally used the search trends data. What type of decisions would you try to make from the data? What type of situations would you find it the most helpful? Are there other types of statistics, for example, that would help you make the same decisions easier or faster?

Cheers,

Scott

Google Keyword Tool Engineer

On January 5th, a PR firm that I do AdWords consulting for, presented me with this problem:

A client of theirs was facing negative press related to alleged corporate malfeasance.   The story had been in the press for about 10 months, but heated up in December when a state attorney general announced he would be filing suit.

The public relations firm asked me to do some basic research into how people are finding the story in search engines.  My first instinct was to check the Google External Keyword Tool to see local search trends for the past 12 months and see which search phrases were getting more searches in December, but I ran into a problem.

I took an initial list of keyword phrases suggested by the client and expanded the list with variations, terms I found in news stories, and my own instincts as to what people would search for.  I entered that list into the Google External Keyword Tool, downloaded the data into Excel, and Voila!

…Except it wasn’t Voila!, it was more like Huh?

When you download keyword suggestions from Google’s Keyword Tool, among the many columns of data are columns for each of the past 12 months with local search volume data for those months. As the screenshot below shows, this “shows a keyword’s fluctuation in traffic over the past 12 months.”

The curious thing was that there was a column for December 2010, but the data was all blank.

At first, I thought it might be might be related to the topic.  Average monthly search volumes were less than 55 searches for most of the 200+ phrases I was analyzing.  Maybe Google is slower to release data for lower volume terms?

But that hypothesis didn’t pan out. Proper names related to the story, which normally get thousands of searches per month, also showed no results for December. Further, I tested known high volume terms (e.g. “wedding favors”) and found the same phenomenon.

I then went back and looked at some prior keyword analysis I’d done, including an analysis from October 14 (for a different client and situation).  Sure enough — September data was blank.

I’m surprised I never noticed this before. Well, it’s possible I had, but it hadn’t really paid much attention because my keyword analysis missions hadn’t been so specific about examining the prior month’s search volumes.

I emailed a contact at Google who is an Account Planner in the AdWords group and told him what I was seeing.  He said we can usually expect the prior month’s data by mid-month.  Based on that, this month appears to be behind schedule.  It’s now January 22, and still no December data.

In an upcoming post, I’ll talk about how I was able to use Google Insights to provide some actionable data without absolute search volume numbers (Insight and Trends only show relative data). Interestingly, both of those tools have shown December data since January 5th, when I first started my analysis.

Updates:

• Google Search Volume Data 27 Days Late?
• How to Use Google Insights for Keyword Research