CPM

CPM is calculated the same simple way everywhere (spend divided by impressions, times one thousand) but what counts as an impression is not identical across platforms, which is the first thing to check before trusting a cross-platform comparison. Google's Display and Video networks generally count an impression when an ad is served to a page, regardless of whether it entered the viewport; Meta counts an impression when the ad is rendered on screen for the user, which is a marginally stricter bar. Neither of those is the same as viewable CPM, a separate, stricter metric (typically requiring 50% of the ad's pixels on screen for at least one second) that some platforms report as an optional column precisely because raw CPM can be inflated by ads served far down a page nobody scrolled to.

CPM interacts most directly with audience size and placement inventory. A campaign targeting a narrow custom audience of past purchasers is bidding against every other advertiser who also wants that same small pool, so it clears at a materially higher CPM than a broad, interest-based audience with millions of eligible users. Placement does the same thing in the other direction: Meta's Audience Network and Google's lower-tier Display inventory carry cheap, abundant impressions, which is why a campaign that expands placements often sees blended CPM fall even while the campaign's real target audience did not get any cheaper to reach. Seasonality moves it too - CPM across most platforms climbs in Q4 as retail demand floods the same auctions, independent of anything an individual advertiser does.

CPM matters most for brand awareness, video view, and reach-style campaigns where the goal is exposure at scale and you are deliberately not optimizing for a click or a conversion event. It matters far less on a direct-response campaign optimizing toward purchases or leads, where CPM is a downstream number the algorithm is managing on your behalf and pushing it down directly (by widening audiences or loosening placements) usually drags conversion rate down with it, netting out to a worse cost per result even as CPM improves.

The common mistake is treating CPM as a cost-efficiency metric to be minimized on every campaign type, when it is really a price-of-audience metric that should be judged relative to what that audience is worth. A retargeting campaign at a $35 CPM reaching people who convert at 8% can be dramatically more efficient than a cold prospecting campaign at an $8 CPM converting at 0.3%, even though the second number looks four times cheaper on its face. The second mistake is comparing CPM across platforms without accounting for the different impression-counting rules described above, which can make one platform look artificially cheaper or more expensive than it actually is.

In reporting, read CPM alongside frequency and audience size, not alone. A CPM that climbs steadily over a campaign's run while the audience size stays fixed is usually a sign of audience saturation - the same eligible pool is getting shown the ad repeatedly, driving frequency up and cost per impression up together, and is a cue to refresh creative or expand the audience rather than a cue to lower the bid. A CPM that jumps suddenly and briefly, without a frequency change, is more often a temporary auction-competition spike worth waiting out.