What is Impression share lost to rank?

A campaign holds 25% impression share, loses 10% to budget and 65% to rank. The money is not the constraint: two thirds of the eligible auctions were lost because the ad was not competitive. Raising the bid addresses one input; raising the expected click-through rate through better ad copy addresses another, costs nothing per click, and lowers the price of the impressions already being won.

This column, which Google Ads and Microsoft Advertising both report beside impression share, estimates the share of eligible auctions an account did not win because its ad was not competitive enough to be shown.

It sits beside the budget column and the two are mutually exclusive explanations. An impression missed because the money ran out is a budget loss. One missed because the ad was outranked is a rank loss. Together with the impression share itself, the three account for the whole eligible pool.

The distinction is the most useful thing in the report, because the remedies have nothing in common. Budget losses are fixed with money. Rank losses are fixed with competitiveness, and money is only one of the ways to buy that.

Ad rank is a combination of the bid and the platform's estimates of quality. The precise formulation is proprietary and differs by platform, and the consistent ingredients are:

The bid. What the advertiser is willing to pay.

Expected click-through rate. How likely the platform thinks this ad is to be clicked in this context, based largely on its history.

Relevance. How well the ad matches the query or the audience it is being shown against.

Landing experience. The speed, usefulness and relevance of the page behind the click.

Because the bid is one input among several, an advertiser can outrank a higher bidder by being more likely to earn the click. This is not a courtesy; it is the platform maximising its own revenue, since it is paid on clicks rather than on impressions.

The practical consequence is the central fact of this metric: three of the four inputs cost nothing per click, and one of them raises the price of every impression the account already wins.

Raising the bid is the intervention people reach for first and the most expensive of the four.

A higher bid applies to every auction, including the ones already being won. In a second-price or similar auction the clearing price is set by the next bidder, so the increase does not automatically raise the cost of existing wins, but it does raise the ceiling at which the account is prepared to transact, and in a competitive set that ceiling gets reached.

The quality inputs behave in the opposite direction. Improving expected click-through raises rank and lowers the bid required for a given position, so it wins more auctions and reduces the price of the ones already won. The same effect applies, more weakly, to relevance and landing experience.

The order that follows is: ad copy first, because it moves expected click-through and relevance together and can be changed today; landing page second, because it is slower but affects everything downstream as well; structure third, where grouping queries more tightly improves relevance mechanically; and bids last, in steps, judged on cost per conversion rather than on position.

In most accounts with heavy rank losses, the fastest improvement is in the match between what was searched for and what the ad says.

An ad group containing thirty loosely related queries served by one generic ad will have mediocre relevance against all of them. The same queries split into tighter groups, each with an ad that repeats the language of its queries, improves relevance and expected click-through simultaneously, with no change in bid.

This is unglamorous work and it is where the returns are. It is also self-reinforcing: a better expected click-through rate begets a better one, because the platform's estimate is built from observed performance.

Not every lost auction should have been won.

An account with a ceiling of forty dollars a conversion, competing in a segment where the winning bids imply sixty, should lose those auctions. The column will report the losses and the losses are correct. Chasing them by bidding above the ceiling converts a healthy account into a busy one.

This is why the column has to be split by segment before it is acted on. A blended figure mixes traffic the account should be winning and losing on price with traffic it is losing on quality, and the two demand opposite responses.

Where the loss is concentrated in segments the business genuinely wants, and the bids required are within the ceiling, the account has a real competitiveness problem worth fixing. Where it is spread across marginal traffic, the right response is frequently to narrow eligibility instead, which raises impression share by shrinking the denominator and concentrates spend where it works.

Heavy rank loss with light budget loss means the account is spending what it has and failing to win. Money is not the constraint and adding it will change very little.

Heavy rank loss with heavy budget loss means both constraints bind. The efficient order is to fix rank first, because a more competitive account converts each subsequent dollar of budget into more volume than it would have done.

Light rank loss with heavy budget loss is the cleanest growth case: competitive, under-funded, and waiting for money.

Light on both means the account is capturing most of what it is eligible for, and further growth requires a larger eligible pool rather than better performance against the current one.

Where a campaign runs on an automated strategy, the bid input is not set by hand, which changes how the column should be read.

The useful reading in an automated campaign is comparative, because the strategy is already declining auctions it does not expect to clear its target — the ad-rank entry sets out why those declines are usually the strategy working rather than a competitiveness failure. Rank losses concentrated in segments where the strategy is meeting its target suggest it is being conservative and the target has room. Rank losses spread evenly while the target is being missed suggest a quality problem the strategy cannot bid its way out of.

The quality half of rank still belongs to the advertiser in either case. No bid strategy writes the ad copy or speeds up the landing page, and both of those raise rank on every auction the strategy enters.

The column moves slowly, because the quality estimates behind it are built from accumulated history rather than from the most recent day.

That lag has a practical implication, set out under Quality Score: a copy change made this week does not show its full effect on rank this week, and the accounts that stay uncompetitive are usually the ones that keep reverting good changes before they register.

Give a quality change at least a couple of weeks and a meaningful volume of impressions before reading the result, change one thing at a time, and keep a note of what was changed and when, so the movement in the column can be attributed to something rather than argued about.

• Read the column against lost-to-budget before acting. The two are mutually exclusive explanations and only one of them is fixed with money.
• Fix the ad copy against the query set first. Relevance between what was searched and what the ad says is the fastest-moving input to rank in most accounts.
• Check the landing page for speed and for whether it answers the query. It contributes to rank and it is usually the least examined half of the account.
• Raise bids last and in steps, watching cost per conversion rather than position. Bidding is the only remedy that makes every impression more expensive.
• Split the column by query or audience segment. A blended figure hides the segments where the account is uncompetitive and the ones where it is fine.