When Does Quality Score Matter?

A few months ago I wrote a post on why you shouldn’t optimize for Quality Score. A few weeks ago I spoke at SMX East on the best ways to optimize for Quality Score in an existing, long running account. I apologized at SMX for changing course and now I am going to change again.

What’s the reason for this QS-bipolarism? I didn’t know until I sat down and did regression analysis on two different types of accounts. The results of it were surprising and not surprising. They were that Quality Score matters and that it doesn’t matter.

Confusing? Yes.

The reason for this is that this analysis showed that for the sample ecommerce accounts Quality Score and CPA are near perfectly correlated. For the sample Education lead gen accounts it is not correlated at all.

Correlation Between Quality Score and CPA Amongst Ecommerce Sample

Correlation Between Quality Score and CPL Amongst Education Sample

On these graphs the blue dots indicate the CPA/CPL for the accounts at each QS level. They exclude brand terms (hence no 10’s) and they are for accounts with identical CPA/CPL goals. (If they weren’t exactly the same the thing charted would have been variance from goal).

The closer the dots to the line the more highly correlated the CPA/CPL is to Quality Score. You can see that ecommerce sample is near perfect on the line and that not only is the Education sample all over the place, there are two QS’s (6 and 4) that have better CPL than the 7 QS keywords.

To deal with the situation I recommend you go account by account and determine if QS and CPA are correlated. You can do your own analysis by using regression modeling, or a simply analysis of CPA by QS will paint a picture for you by simply graphing out the CPA by QS on a graph and seeing if the patter in linear. If they are then spend your time optimizing for QS. If they are not correlated than you time might be better spent elsewhere.