« Adinton Blog

Adwords Quality Score Demystified

Let me remember, what Adwords Quality Score is:

Quality score (QS) is the metric that google uses to evaluate the “usefulness” of a keyword. This, together with the bid offered at the auction is going to determine the adrank of this keyword and thus the position of the ad in the google search page. By “usefulness” we mean that google choses the keyword that is going to both satisfy the person searching in their page but also that is going to lead to a conversion. This second statement we will study further ahead in this report.

The variables that google uses to evaluate the QS are very diverse and the formula is extremely complicated, but we can still try to find its main contributors and, to some extent, the weight of these contributions. The main component of QS is by far the CTR. Google charges per click and not per impressions, so they are interested in advertisers that have the highest CTR possible. Fonts from google stablish this contribution to around 60%. But google is also interested in providing a good service and making its users happy, so it tries to make sure that the people who click on the ads have a good experience. This means that CTR is not the whole deal. There have to be many more indicators which ensure that the user not only clicks on the ad but also finds what he or she is looking for.

In the following report we attempt to study the effect that CTR has on QS, as well as how related are the conversion rate (CR) and the superconversion rate (SCR) and the QS. The superconversion rate is defined as SCR=CTR*CR and gives us the conversions per impression of the keyword, which, after all, is what both google and the advertiser care about.

We have studied the data from 6 different companies during a span of between 6 and 10 years and drawn some very interesting conclusions about what the QS is and what we can do about it. It’s important to note before starting that the data we have is not complete, nor accurate, because some of the companies didn’t do a very good job in tracking their conversions. Nevertheless we believe that we can weed out the bad data and get reliable conclusions from what’s left.

So, what we did is take all the non-brand keywords that have more than 100 clicks and then we have averaged the CTR, CR and SCR for every QS, plotted them and found regression lines to get an idea of their correlation.

This is the correlation we analyzed with one of our clients:


In this plot we can see that the R2 of the CTR line is of 0.79, so not a bad agreement. The R2 of the conversion and the superconversion rates is also quite good. You will notice that in this plot there are no keywords for QS=10; this is because the data that we had for this value was incomplete and untrustworthy. We’ll see that this happens throughout the study.


As we can see the agreement in this case is a bit worse. We are unsure at this point what is the cause of this, but the correlation is still quite decent so we’ll leave it as it is.


These results are similar to the ones from Client#2, so we can draw the same conclusions as before.


We can see that in this case the correlation is very good for the CTR but is a bit off for the conversion rate. We believe that this is caused by the fact that the data is not particuarly good for the conversions. Nevertheless the SCR regression line is also very good.

Now we will go to the average of 5 companies, 4 of them presented before.


We can see here that the agreement for the CTR is extremely good. This means that, as expected, the quality score is highly influenced by the CTR. From this diagram we also see that the Conversion rate is really off using a linear trend line, so what we have done in the next plot is to adjunt an exponential trend line as well as removed the QS=10 points because of bad data. The result is the following:


In this plot we can see the extraordinary agreement that there is between the visible QS and the variable that we are studying.

From this we gather again that the QS is very strongly related to the CTR, and further we see that the relation is not linear, but instead google rewards only the extremely high CTR’s with the top QS. But what’s more extraordinary is that the relation between QS and SCR is even better than the one between QS and CTR. As far as we know google doesn’t use the information about the conversions to calculate the quality score. This means that the multitude of variables that google uses to compute QS are in fact helping you to convert more.

So we can see that the QS strategy is helping you improve your marketing strategy, is helping google make more money and is keeping the user happy, so that it’s a win-win-win strategy. However, one musn’t drive all their efforts only at having the maximum possible QS; there can be keywords that are very profitable albeit having a low QS of vice-versa, so we also need to keep track on that.


Leave a reply

Your email address will not be published. Required fields are marked *


Log in with your credentials

Forgot your details?