A DTC home goods account spending roughly $35K a month on Meta paid 19% more per thousand impressions in June than it did in January, $16.40 against $13.80. Its CPA never left the $41 to $44 band, and June revenue per Meta dollar finished 6% higher than January. The auction did not get kinder. The account outran it, because the two variables an advertiser controls, creative and the landing page, improved faster than the one it does not.
The auction got more expensive for almost everyone
Meta CPM inflation in 2026 is portfolio-wide, not an account-level anomaly. Threadpoint reports its portfolio-weighted average Meta CPM up 13% year over year, from $15.07 to $17.01, measured January through May across 17 accounts 1, and the same post cites a Ryze benchmark with average Meta CPMs up 20%, from $11.82 to $14.19 1. If your CPMs are up double digits this year, the market moved, not just your ad sets.
One counter-current is worth holding onto. Instagram CPMs declined 3% in Q1 2026, the first decline since 2023, while Reels reached 33% of Instagram impressions, per Tinuiti 2. Impression supply is expanding where short vertical video runs, and expanding supply is the one force pushing back on auction prices. That detail shapes what the account in this piece did with its creative mix.
The account: rising CPMs, a CPA that refused to move
The pattern comes from a DTC ecommerce account in home goods, spending in the $35K a month range on Meta, and the numbers are the firm’s own anonymized figures. From January to June 2026, account CPMs rose 19%, from $13.80 to $16.40, in line with the market data above. CPA held between $41 and $44 the whole stretch, no month outside the band.
The account did not hold flat by shrinking. Spend stayed level, and June revenue per Meta dollar came in 6% above January. So the account paid the market’s inflated price for attention and converted that attention well enough to come out ahead of where it started.
Just as telling is what did not change. Nobody touched the bid strategy. Nobody restructured audiences or layered in exclusions to chase a cheaper pocket of the auction. Targeting stayed broad the entire six months, because narrowing it would have traded a visible CPM problem for an invisible learning problem. Every hour that could have gone into settings went into the two inputs that show up in the denominator of the CPA equation instead.
CPM is the auction’s number; CTR times CVR is yours
With Meta CPMs rising, how to keep CPA flat becomes an arithmetic question rather than a bidding question. CPA is CPM divided by the product of click-through rate and conversion rate, scaled per thousand impressions. The auction sets the numerator, and it is set by every other advertiser’s budget as much as by yours. The denominator, how often the impression becomes a click and how often the click becomes a purchase, belongs to you.
Run the arithmetic and the case stops looking lucky. A 19% CPM increase is fully offset by a roughly 19% improvement in click-to-purchase efficiency. This account got a 48% conversion rate improvement from the page alone, 2.3% to 3.4%, which covered the inflation with margin to spare. That margin is where the 6% gain in revenue per Meta dollar came from.
What changed: ten assets a month and a rebuilt product page
Creative production went from four new assets a month to ten. The mix was statics plus 15-second vertical video cut for Reels, which is where the impression supply is growing, and everything ran on a weekly kill-and-replace cadence: anything past its fatigue point came out, something new went in. That cadence is not a style preference. It follows from fatigue windows compressing to five to seven days, the pattern we documented in creative fatigue hits in five days now, and the production calendar this account ran was built to stay ahead of exactly that window.
The page work was just as unglamorous. The product page for the two hero SKUs was rebuilt: mobile load time cut from 4.1 seconds to 1.9, the above-the-fold offer rewritten to match the ad hook that brought the click, and the review count surfaced next to the CTA instead of buried below the fold. Landing page conversion rate went from 2.3% to 3.4%. Nothing on that list is exotic, which is the point we argued in why the landing page matters more than your bid strategy: the page is where a paid click either becomes revenue or becomes a CPM you paid for nothing.
Auction inflation is a tax on static accounts
When CPMs inflate, the two most common responses make the problem worse. Tightening targeting shrinks the audience the delivery system can learn from, and cutting spend shrinks the conversion signal that trains it; both raise the effective cost of the next conversion while the operator waits for prices to come back down. Prices are not coming back down. Accounts that respond with creative velocity and page work hold CPA through the same inflation that wrecks their neighbors.
This case is the position on our /about page in miniature: the platforms have automated the targeting layer, so the leverage sits in creative, offers, feeds, and landing pages. It is also why the way we structure engagements separates media management from creative and page work. The hours go to the parts of the equation the auction cannot touch, because those are the only parts that moved this account’s number.
CPM is rent, and rent in this neighborhood goes up most years. You do not negotiate with it; you out-earn it. The accounts still shipping ten assets a month and rebuilding pages hold their CPA while everyone else tightens targeting and waits. Auction inflation is a tax on static accounts, and staying static is a choice.