Rising impression share isn't always a win

Impression share is the percentage of eligible auctions your ad actually appeared in. If there were 1,000 searches or feed loads where your ad was eligible to compete, and your ad showed up in 700 of them, your impression share is 70%. Both Google and Meta report versions of this metric at the campaign or ad set level, and it's one of the first numbers most reporting dashboards surface, because it reads as a clean scoreboard: bigger number, bigger share of the pie you're competing for.

That framing invites an obvious conclusion. If impression share climbs from 40% to 65% over a month, the natural read is "we're winning more of the auctions we're eligible for, so we're reaching more of our potential audience." For a founder who's 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 it's measuring effort translating into results. It usually is. But the metric can't tell you, on its own, which side of the fraction actually moved.

Impression share is a ratio: impressions you got, divided by impressions you were eligible for. A ratio rises if the top number goes up, but it also rises if the bottom number goes down, even if the top number stays flat or falls. Both platforms are explicit that impression share is calculated against "eligible" auctions, not total possible auctions or total market volume, and eligibility itself is not fixed. It moves with your own bidding and budget settings.

This is the mechanism that breaks the intuitive read. Automated bidding systems, whether that's a Target CPA, Target ROAS, or similar bid strategy on Google, or an equivalent cost-control setting on Meta, are built to hit a cost or return target, not to maximize reach. When you tighten that target, asking the system to hold a lower cost per acquisition or a higher return, the bidding algorithm doesn't try harder in every auction. It becomes more selective. It voluntarily sits out auctions it forecasts as too expensive relative to your new target, because bidding aggressively in a pricey auction would blow the target it's being held to. Each auction it exits stops counting as an auction you were even eligible to compete in for that data point. The eligible pool shrinks. And because impression share is impressions won divided by that shrinking pool, the percentage can go up even while the campaign is reaching fewer people than before.

Here's a hypothetical to make the mechanism concrete. Imagine a Search campaign running with a Target CPA bid strategy. The team, trying to improve efficiency, lowers the target CPA, asking the algorithm to deliver conversions at a lower cost per conversion than before. Over the following few weeks, impression share on that campaign climbs from around 35% to around 60%. On a dashboard, that's a clean upward line, and it gets reported as a win: "we're capturing almost twice the share of available auctions we were a month ago."

But in this hypothetical, raw impressions and clicks over that same stretch actually fell. What happened is that the tighter CPA target pushed the bidding system to stop competing in a large slice of auctions it used to enter, specifically the more expensive, more competitive ones where hitting the new lower cost target wasn't realistic. It kept winning most of the auctions in the smaller, cheaper pool it was still willing to enter, and won a smaller absolute number of them than before. The percentage rose because the denominator shrank faster than the numerator did. Nothing about the campaign's actual reach improved. It got narrower and, on this one metric, better-looking at the same time.

Impression share is only interpretable next to the absolute numbers underneath it: raw impression count and raw click count over the same period. Those two numbers together tell you which side of the fraction moved.

If impression share is rising while raw impressions and clicks are flat or falling, that's a shrinking pool, not real progress. The campaign is competing in fewer auctions and winning a larger share of a smaller set, which is a sign the bidding system pulled back, not that it got more competitive. If impression share is rising while raw impressions and clicks are flat or rising, that's the genuine version of the story the dashboard implied: the campaign is winning more of a pool that stayed the same size or grew, which does mean more real reach. Same headline metric, opposite underlying reality, and the only way to tell them apart is to look at the volume numbers sitting next to it.

If you manage an agency or oversee someone else running your paid media, rather than touching the platforms yourself, impression share is an unusually easy number to report upward without it reflecting real growth. It requires no fabrication, just a tightened bid target and a few weeks, and it produces a chart that looks like unambiguous progress to anyone who doesn't know to ask what moved underneath it.

The fix is a standing question, not a one-time audit: whenever an impression-share improvement is presented to you, ask to see the absolute impression and click trend from the same period, on the same campaign. If both are rising or holding steady alongside the share increase, that's a real result. If they're flat or declining while impression share climbs, the honest description of what happened is that the campaign got more selective about which auctions it entered, not that it started winning more of a stable, sized audience. Either outcome might be the right call strategically, but only one of them is the win the chart implies, and you can't tell which without the raw numbers.