How the three quality components combine into one score

You meet it on the Keywords page under the Campaigns menu, once you open the columns icon and add the Quality score columns. Google shows four of them together: Quality Score, Ad Relevance, Landing Page Exp. and Exp. CTR, each of the last three reading Above average, Average or Below average. There are historical variants of all four for past reporting periods.

The question follows immediately. Two keywords both read Below average on one component and Average on the others, and they carry different scores, so you start trying to work out the arithmetic.

Google states that Quality Score is calculated from the combined performance of expected click through rate, ad relevance and landing page experience, and that each is evaluated as Above average, Average or Below average against other advertisers whose ads showed for the exact same search over the last 90 days. It does not publish how the three combine, so the mapping has to be observed rather than derived.

Grouped by score, the whole table is short. Google labels each component Above average, Average or Below average, and the triples below run in the order Ad relevance, Landing page experience, Expected CTR, shortened to above, average and below. One combination reaches 10: all three above average. One gives 9: average ad relevance with both others above. Three give 8: above, above, average; above, average, above; and below, above, above. Five give 7: above, above, below; above, average, average; above, below, above; average, above, average; and average, average, above. Five give 6: average, above, below; average, average, average; average, below, above; below, above, average; and below, average, above. Five give 5: above, average, below; above, below, average; below, above, below; below, average, average; and below, below, above. Two give 4: average, average, below and average, below, average. Three give 3: above, below, below; below, average, below; and below, below, average. One gives 2: average, below, below. One gives 1: all three below average.

Read in that order, two things fall out of the table, and both are observations of the data rather than claims about Google's formula. The mapping is not additive: below, above, above scores 8, where treating the three statuses as equal steps would predict 7. And the components are not interchangeable: being strong on Expected CTR alone, below, below, above, scores 5, while being strong on Ad relevance alone, above, below, below, scores 3.

The cost of not knowing this is effort spent on the wrong component. If the three are assumed to be equal, the obvious move is to work on whichever one is easiest to change. The table says the three are not equal, so the easiest change and the largest change are not necessarily the same one.

Because every cell carries zero variance, the three component statuses fully determine the Quality Score integer in this data. The score therefore adds no information beyond the three labels already on the screen. That is the practical conclusion: read the components, not the number, and spend the time on whichever one reads Below average.

It is not Ad Rank, and it is not a lever you bid with. Google states that Quality Score is an aggregated estimate of your overall performance in ad auctions, and is not used at auction time to determine Ad Rank. Whether the three underlying components affect what you pay is a separate question, and this page does not answer it.

It is not a reconstruction of Google's formula either. Nobody outside Google has published how the three components are weighted, and this table does not reveal it. It records which integer appeared for each combination in one set of accounts, which is a different and more modest claim.