Today we have a guest post from one of our PPC Hero allies! Craig Danuloff is doing a ‘Blog Tour’ to promote his upcoming book ‘Quality Score in High Resolution‘ which will be released in May and is now available for discounted pre-order. In this post he explains some of the lesser-known facts of the quality score metric reported by AdWords


Ready for a simple PPC question?

True or False: Quality Score is a number between one and ten that is assigned to every keyword in your AdWords account.

<Play that Jeopardy music in your head and decide.>

Got your answer?

Sorry, trick question. The answer depends on which quality score you mean.

There is a metric named ‘quality score’ that appears for every keyword in your account. In terms of that quality score, the statement above is true.

But there are other ‘quality scores’ that are used inside the AdWords system to determine whether or not your ads are eligible for auctions, which position they appear in, and even how much you pay per click. The above statements are not true for those versions of quality score.

I’ve taken to calling the first quality score, the one that is a number between one and ten and is assigned at the keyword level, visible quality score. Because you can see it.

More About Visible Quality Score

Visible quality score is a simplified proxy for the average ‘real’ quality scores for a keyword. It is based on a simpler formula than the one used by the real quality scores. It’s also expressed in a different numerical range and distribution.

Visible quality score gives you an idea of how the keyword is performing in terms of quality score. But it doesn’t consider all of the factors that go into the more critical quality score values, so there are things it can’t tell you.

One of the reasons why quality score is so confusing is that people try to answer a lot of the questions they have about quality score, or understand the quality score issues in their account, based on what they see in the visible quality scores. But in many cases visible quality scores lack the level of detail necessary to answer or explain those issues.

Google exacerbates the issue by calling two different things by the same name. Suppose you read a help file that says ‘quality score considers the relevance of your search queries to your keywords’ and so you decide to review query performance in your account. But when you look at your search query reports, and then look at the column in your reports that is named ‘quality score’ it isn’t clear that the quality score you can see isn’t the one that is effected by search queries. It’s easy to get frustrated or confused.

Here’s a few facts that will help you to understand the limits of visible quality score:

  • It only considers performance of search queries that are identical to the keyword. If a broad match keyword was shown in response to 100 different search queries, but only 5 of them were identical to the keyword, the visible quality score only reflects the CTR from those five instances. Match type never matters, just if the query is identical to the keyword.
  • There is only one visible quality score per instance of a keyword in your account. If you buy the keyword three times in three different match types, the aggregate performance from instances where the query was identical to that keyword is used to calculate a single visible quality score that is shown for each instance of that keyword. If you put that keyword in 10 different geographically targeted campaigns, again the aggregate performance of all instances is rolled up to a single visible quality score that is shown for each of them.
  • Visible quality score also ignores the geography of the searchers. This should be clear from the prior example – the keyword will get the same visible quality score in a geo-targeted campaign that has a high CTR as it will in another geo-targeted campaign where it has a low CTR.
  • Visible quality score blends the performance of all text ad creatives and display URLs. The various text ads in your ad groups often earn a broad range of different quality scores, ad copy can be edited or new versions introduced at any time. The impact of these on CTR are all blended into the visible quality score for any given keyword.

Working Beyond Visible Quality Score

Visible quality score has limitations, but it is useful for understanding how your keywords are performing and for managing and improving quality score. It is highly correlated with CTR (another metric where what is reported is an average giving less detail than we might want) which isn’t surprising given that’s effectively what it is tracking.

Visible quality score provides a great warning system. Since visible quality score only reflects queries that are identical to the search query, it probably reflects a relatively positive view of actual quality scores. So if it’s a low number – anything below 6 – you should worry and take corrective action.

There are at least three main issues that visible quality score doesn’t reflect that you should monitor yourself:

  • Search Query Performance – Create new keywords from high volume or important search queries, so they get their own visible quality scores. This is the only way to see how they’re really performing.
  • Geographic Performance – This is a tougher one but if you geo-target campaigns, you can compare CTRs for the same keywords in different geographies. Any areas where a keyword is significantly under-performing may be weighing down visible quality score for that keyword.
  • Ad copy differences – When one text ad underperforms other ads in the same ad group in terms of CTR, it can significantly weigh down visible quality scores. Writing and testing a lot of ad copy versions is always important, but don’t neglect the step where you pause the losers. Letting them run once they’ve proven statistically significantly worse is harmful.

The quality score calculations Google performs to determine eligibility, ad rank, and CPC are complex and done in real time. It would be very difficult for Google to share this data with advertisers.

Visible quality score is not a bad compromise, but like any summary it lacks some of the important details contained in the raw data. By understanding the limitations you can get the most out of visible quality score but also know when it’s time to look past this metric to other clues to tune and improve the performance of your account.


Craig Danuloff is founder and president of ClickEquations, and author of ‘Quality Score in High Resolution‘. The book is scheduled for release in June 2011, but discount pre-orders are available until May 26, 2011.