01 Apr Insider scoop of the year: AdWords Quality Score algo exposed
A former Google employee (who must remain of course anonymous) unveiled us the original algorithm used for Quality Score calculation in AdWords.
Our source, who we will call April F., explained us that the quality score not only sets the cost per click (the higher the QS is, the lower the CPC), but does a lot more. It determinates if our ads will participate to auctions or fixes the estimate for both first page and premium positions on the top of SERPs.
The algo was revised last October for the first time since 2005, officially including extensions click through rates. April F. told us that in the search network the Quality Score is influenced by:
- keywords CTR in Google (main factor)
- account past general performances
- historical CTR of display URLs
- user experience on the landing page (bounce rate & time on site)
- keywords pertinence to ad texts
- keywords & ads pertinence to landing page texts
- keywords relevance for the search query
- performance in the geographical area where ads are shown
- device used to display ads
- other minor factors of relevance.
Display network QS calculation is different and it is not visible anywhere in the control panels. April F. told us that, except in the case of specific placings selection, the targeting is only indirectly adjustable by advertisers, and so the main factors are quite different:
- keywords & ads pertinence to the site where they should be shown
- performances in similar sites where ads were shown previously
- landing page quality factors (such as transparency and clearness of both interfaces and business models)
- other minor factors of relevance
Besides April F. revelations, we wanted to fully unravel the secrets of all other aspects of this important matter, which until now have always been surrounded by an aura of mystery. After three years of efforts (leaving to Google several months of our earnings) we made it. We finally found all the secret ingredients of the original Quality Score recipe.
Ingredients for no limits in-target users:
- a bounce of keywords (not too many for each group)
- the right corresponcence types (the lower the QS, the stricter the matches)
- several ad variants for each group (with regular tests)
- a lot of spicy extensions
- abundant negative keywords
- 1 unlimited spending credit card
- several competitors benchmarks
- a pinch of fantasy
- some years of experience
- a bit of courage and recknessless
The compound should assume a fairy green and conversions rich appeareance. If the mixture turns red means that you have to continue to test different proportions among ingredients and, may be, a different structure pot.
Not all donuts come out first with a hole right in the middle.