Rising impression share isn't always a win

Impression share is the impressions your ads received divided by the impressions Google estimates they were eligible to receive. Google writes it as a formula: “Impression share = impressions/total eligible impressions”. If Google estimated you were eligible for 1,000 impressions and your ads took 700 of them, your impression share is 70%. It is one of the first numbers a reporting dashboard surfaces, because it reads as a clean scoreboard: bigger number, bigger share of the pie you are competing for. Everything below is Google Ads, whose documentation defines both the column and the estimate sitting underneath it.

That framing invites an obvious conclusion. If impression share climbs from, say, 40% to 65% over a month, the natural read is that you are winning more of what you were eligible for, so you are reaching more of your potential audience. For a founder who has never run a campaign, an agency client skimming a report, or anyone who tried ads once and got burned, that reading feels safe precisely because it sounds like effort turning into results. Sometimes it is. But the metric cannot tell you, on its own, which side of the fraction moved.

A ratio rises when the top number goes up. It also rises when the bottom number goes down, even while the top number is flat or falling. So the whole question is what the bottom number is, and it is not a count of your market. Google says “Eligible impressions are estimated using many factors, including targeting settings, approval statuses and quality.” It is not a count of every auction that happened either. Google scopes it to where you were in contention: “Impression share includes all auctions where your ad showed, and all auctions where your ad is competitive enough to show.” The example it gives of that boundary is measured in bids. The estimate “could include auctions where your ad could show at twice its current bid, but could exclude auctions where your ad is estimated to need a 1000% bid increase in order to appear”.

Then it says the boundary moves. “Changes to your bids, quality or Google's ad systems may change the set of auctions in which the system estimates that you were competitive.” That is the sentence that makes this metric hard to read. The bottom of the fraction is not a fixed market you are being scored against. It is Google's estimate of where you were competitive, and Google says your own bidding may change it. It adds a caution of its own in the same paragraph, about the noise rather than the mechanism: “Small fluctuations over time don’t necessarily indicate that action is needed.”

Here is a hypothetical to make that concrete, using the lever Google names most plainly. A Search campaign runs across a whole country. To cut waste, the team narrows it to the regions that convert best. Over the following weeks impression share on that campaign climbs, and on a dashboard that is a clean upward line, reported as a win: we are capturing a far larger share of the impressions available to us.

Every word of that is true, and the last four are doing the damage. The impressions available to it are a smaller set now, by the same edit, and Google's help page names that directly: “If you decrease your regional targets, you may see an increase in impression share. But keep in mind that decreasing regional targets reduces the potential number of impressions available to you.” Raw impressions and clicks over the same stretch can fall while the percentage rises. Nothing about the campaign's reach improved. It got narrower and, on this one metric, better looking at the same time. Whether narrowing was the right call is a real question, and it is a different question from whether the chart shows progress.

The version of this story you will hear most often involves a bid strategy rather than a map. A team tightens a Target CPA or raises a Target ROAS, impression share climbs afterwards, and the explanation offered is that the strategy is now sitting out the auctions it cannot afford, so the pool underneath the metric shrank.

That a target change moves bids is not folklore. Google says “Google Ads uses your target CPA to set a bid based on the likelihood of the ad to convert.” Target CPA and Target ROAS are both Smart Bidding strategies, in Google's own grouping: “Target CPA , Target ROAS , Maximise conversions and Maximise conversion value are all Smart Bidding strategies.” So the bid moving is not a side effect of changing a target, it is how the strategy works, and Google has already said that changes to your bids may change the set of auctions its estimate is built from. Both halves of the fraction are in play at once.

Google also documents an impression share column going wrong under a bid strategy. Of the Lost IS (budget) column with Maximise conversions, it says “the column is incompatible with the bid strategy”. Its reason covers both of the maximise strategies, which it says “are designed to spend the full daily budget”, while the column “uses a different definition to account for lost impressions”. Different column, different strategies, but it is the same lesson: these metrics and these strategies interact, and Google will sometimes tell you where.

What Google publishes about moving a target is written in terms of volume. On Target ROAS it says “setting a target that's too high may limit the amount of traffic that your ads may get”, and, under value-based bidding, “When using value-based bidding with a Target ROAS, if you want to increase conversion volume, consider gradually reducing the Target ROAS.” with the note that “This allows the bid strategy to enter more auctions and generate more volume.” On Target CPA it says “Setting a target that's too low, for example, may cause you to forgo clicks that could result in conversions, resulting in fewer total conversions.” Traffic, auctions entered, clicks, conversions. That is the vocabulary, and none of it is a quantity you can subtract from a percentage.

So the honest reading of a rise in impression share after a target change is that the top and the bottom of the fraction can both have moved, and that the percentage on its own cannot tell you which. The good news is that you do not need the mechanism, because the arithmetic is available to you.

Google's formula rearranges. If impression share is impressions divided by eligible impressions, then eligible impressions are impressions divided by impression share. Take the raw impression count for a period, divide it by that period's impression share written as a decimal, since 70% is 0.7, and you have the pool implied by Google's own formula. Do it for this month and last month and the question stops being interpretive. One number went up or it went down.

That recovers the estimate only where the impressions you divide and the share you divide by describe the same traffic, and Google names several places where they do not. Search partner traffic is out of one and not the other: “Search partners doesn’t distinguish between 'top' vs 'others' and won’t be included in impression share data.” Campaign types do not combine: “Impression share metrics are reported separately for each campaign type and aren’t aggregated across the account.” On Shopping campaigns the impressions column is not the number the ratio was built from at all, because “only your first impression for each search will qualify for impression share”. Performance Max has its own numerator, “Performance Max impression share = impressions from Search + Shopping / total eligible impressions from Search + Shopping”. And Hotel has its own basis entirely: “Impression share for Hotel campaigns is calculated on where your ad appears in the Hotel ads booking module.” So do the division inside a single campaign type, on a Search or Display campaign, against an impression count that does not include search partners. Anywhere else, the two columns are describing different things and the quotient is not a pool.

The division matters because the rule most people carry instead is wrong, and it is wrong in the flattering direction. The usual version is that a rising share is real as long as impressions held steady. It is not. If impressions are flat and the share rose, the pool must have fallen, by the definition of the ratio. Rising impressions are not enough on their own either. If impressions rise by 5% while the share is multiplied by 1.2, say from 40% to 48%, the implied pool fell by an eighth. For the pool to have grown, impressions have to rise by a larger proportion than the share did, and that is a comparison, not a glance.

Two caveats on the number you get, both of which follow from what Google says about it. The bottom of the ratio is an estimate, so treat the result as a direction rather than a measurement, and take Google's own advice not to read small movements as a signal. And a pool can shrink for a reason you chose, which is not the same as a reason to regret. The point was never that shrinking is bad. It is that a percentage on its own cannot tell you it happened.

If someone else runs your paid media, impression share is an unusually easy number to put in a report without it standing for growth. No figure has to be wrong for that to happen. A narrower target area and a few weeks will do it, and Google's own help page is where you can read that the share going up and the impressions available going down can be the same edit.

The fix is a standing question rather than a one-off audit. Whenever an impression share improvement is put in front of you, ask for the raw impression count on the same campaign over the same two periods, and do the division yourself. Then ask what was edited in that window. Google's general tips for raising impression share are a short list to check the answer against: increase the campaign budget, increase the bid, decrease regional targets, re-adjust the budget when regional targets grow, and improve ad quality on Search Network campaigns. For Display Network campaigns it names improving display ad quality and adjusting managed placements, tells you to re-adjust the budget if you add placements, and warns that cutting them may raise impression share while the size of the pie has been reduced. If the implied pool held or grew while the share rose, the chart is telling the truth. If the pool shrank, the honest description is that the campaign got narrower, which may well have been the right decision, and is a different sentence from the one the chart is making.