What is Ad Rank?

Two advertisers compete. A bids A$3.00 and its quality signals multiply to an index of 4, giving a rank of 12. B bids A$5.00 with an index of 2, giving 10. A wins on the lower bid. A then pays only what it needed to just beat B's rank of 10: 10 ÷ 4 = A$2.50, and one increment more of whatever the auction works in. The better-quality advertiser wins the position for a shade over A$2.50 against a competitor willing to pay A$5.00.

Ad Rank is the value Google Ads computes for each competing ad in each auction. It decides whether an ad shows at all, in what position, and what it pays. Every other search metric an advertiser watches is downstream of it.

Its defining property is that it is not the bid. The bid is one factor, multiplied by several estimates of quality, and the product is what gets ranked.

Ad Rank ≈ bid × expected CTR × ad relevance × landing page experience

The exact formulation is proprietary, includes the expected impact of formats and extensions, varies by context, and changes over time. What is stable and worth building an account around is the shape: money multiplied by quality, not money alone.

The reasoning is commercial rather than editorial.

A search platform is paid when somebody clicks. An impression given to an ad nobody clicks earns nothing. So the value of showing a particular ad is the bid multiplied by the probability of a click, and ranking on that product maximises the platform's revenue per impression.

That the arrangement also rewards relevant advertising is a consequence rather than a motive, and it is a reliable one. An advertiser cannot talk the platform into a better rank; it can only make an ad more likely to be clicked.

The same logic explains the thresholds. An ad whose rank falls below a minimum is not shown at any position, because showing it would waste an impression the platform could have monetised. A high bid does not rescue an ad with weak quality signals, and advertisers who try to buy their way past a threshold discover this expensively.

The clearing price follows from the ranking, and it is where the quality half pays for itself twice.

To hold a position, an advertiser needs a rank that just beats the competitor below. The bid required is that competitor's rank divided by their own quality estimate, and then the smallest increment the auction works in on top of that. Stronger quality means a smaller bid buys the same rank.

Work through the numbers in the example above. A bids A$3.00 with quality index 4, for a rank of 12. B bids A$5.00 with index 2, for 10. A wins. To reach B's rank of 10, A needs a bid of 10 ÷ 4 = A$2.50; to beat it, one increment more. So A pays a shade over A$2.50 — around half what B was willing to pay, for the better position.

This is the mechanism behind the claim that better creative lowers costs. It is not a general principle about advertising; it is arithmetic in the auction.

Advertisers do not see Ad Rank. What they see are its consequences, and three reports carry most of the signal.

Impression share lost to rank estimates how much of the eligible market was lost because rank was insufficient. This is the direct read.

Quality Score and its three components report the estimated quality half, periodically and rounded. It is a diagnostic rather than the live input, and the components matter far more than the number.

Average position indicators — the top-of-page and absolute-top impression share figures — say where the ad appeared when it did, which is a different question from whether it appeared.

Read together they separate the two reasons an account is small: not competitive enough, or not funded enough. The remedies have nothing in common.

Because the bid is the only input that raises costs while raising wins, it belongs last.

Ad copy first. It moves expected click-through rate and relevance together, it can be changed today, and it is the component with the most weight.

Structure second. Tighter groups let each ad speak to the queries it serves. Relevance is largely a proximity problem and structure is how proximity is arranged.

Landing page third. Slower to change, and it affects rank as well as everything after the click. Frequently owned by somebody outside the advertising team, which is why it is the component most often left alone.

Bids last, in steps, judged on cost per conversion rather than on position.

Under an automated bid strategy the last step is not available in the same form — the strategy sets the bid from its target — which makes the first three the whole of the advertiser's contribution to rank, and raises rather than lowers their importance.

Losing is frequently correct. An account with a ceiling on what a conversion can cost should lose the auctions that clear above it. Rank losses concentrated in traffic the business cannot afford are the strategy working, not a problem. Splitting the lost-to-rank column by segment is what separates the two cases.

Quality signals are slow. They accumulate from observed behaviour, so an improvement made this week shows over the following weeks. Changing an ad, reading the numbers three days later and reverting is how accounts stay where they are.

The quality half is not only the three headline components. The expected impact of ad formats — sitelinks, callouts, structured snippets, images, whatever the surface supports — enters the calculation as well.

The practical effect is that adding relevant extensions can raise rank without any change to the bid, the copy or the page. It is among the cheapest available improvements in a search account, it is frequently skipped because it feels cosmetic, and it compounds with everything else because a higher rank also makes more formats eligible to show.

The qualifier is relevance. Extensions that do not match what the query wants can be ignored or can dilute the ad, and a long list of generic callouts is not the same as two that answer the question.

Where a campaign runs an automated strategy, the bid half of the calculation is no longer set by hand, and this changes what an advertiser can do about rank.

The strategy bids up to the level its target implies and no further. Rank losses under that arrangement are frequently the strategy behaving correctly: it declined auctions it did not expect to clear the target. Loosening the target to chase rank spends money the target existed to protect.

What does not change is the quality half. No bid strategy writes the ad, tightens the structure or speeds up the page, and all three raise rank on every auction the strategy enters. Under automation they become the entirety of the advertiser's influence over whether the ad appears, which is an argument for spending more time on them rather than less.

The minimum rank an ad must clear to appear is not a single number. It varies with the query, the position on the page, the device, the context of the search and the nature of what is being advertised.

The practical consequence is that an ad can be eligible in one context and not another with no change on the advertiser's side, and that a campaign appearing sporadically is frequently sitting just at a threshold rather than losing on price. Raising the bid moves it over the line for a while; raising the quality signals moves it over the line durably and more cheaply.

• Raise the expected click-through rate before the bid. It is the only input that wins more auctions and lowers the clearing price at the same time.
• Tighten the match between query, ad and landing page. Relevance enters the calculation directly and responds to structure faster than to rewriting.
• Fix the landing page for speed and for whether it answers the query. It is an input to rank, not just to conversion rate.
• Check the lost-to-rank column by segment rather than in aggregate, so the work goes where the account is genuinely uncompetitive rather than where it is correctly declining.
• Raise the bid last and in steps, judged on cost per conversion rather than on position.