Meta CPMs are rising into H2. Protect your payback

The ads you buy on Meta are quietly getting more expensive, and your cost to win a new customer is rising with them, even if you have not changed a thing. Across 2025, the price to show your ad to a thousand people (Meta calls this CPM) rose roughly 20% by one large benchmark, and Meta's own results show the average price per ad climbing every quarter. Into the back half of 2026, holiday demand and a more crowded auction push the same ad space higher again. The reflex is to cut spend. That protects nothing. A pricier ad is a cost-per-customer problem, and the levers that fix it are better ads, a stronger offer, and customers who stay longer. Not a smaller budget.

What happened

Meta got more expensive, and the data agrees from every side. Triple Whale's benchmark report, covering close to 35,000 brands over 2025, found the price per thousand ad views up 20% year over year at a median near $13.48. The share of people clicking rose 13.5%, and the cost per actual result stayed close to flat. In plain terms: showing the ad got pricier, but sharper targeting and fresher ads kept the cost of a real sale from running away. Meta's own full-year 2025 results tell the same story from the supply side: average price per ad up 9% for the year on 12% more ad slots. More space for sale and a higher price at the same time happens only when advertiser demand grows faster than the supply.

Then there is the calendar. The holiday quarter is the most expensive stretch every year, because every retailer floods the auction at once. One analysis put late-2025 prices about a quarter above the annual average, with Black Friday week two to three times normal. Expect a steady climb through the summer, then a sharp step up around the holidays.

The money angle

The price per thousand views is the raw material cost of your whole ad operation. You pay for views. Some share of viewers click. Some share of those click-throughs buy. Divide your spend by the number of buyers and you get your cost per new customer. When the price of views rises 20% and nothing else moves, your cost per customer rises about 20% too. On a thin-margin business, that can flip advertising from profitable to money-losing overnight.

So the question is never how to pay less per view. The auction sets that price, and you are one bidder among millions. The real question is how to get more customers out of each expensive view, and how to make each customer worth more. Three numbers do that, and you control all three: the share of visitors who buy, the size of the average order, and how much a customer spends with you over the years. Lift any one and a higher ad price stops hurting. Cutting spend changes none of them. You still pay the higher price per view, just on fewer views. Revenue shrinks and the math stays broken.

Where it breaks

The quiet leak is treating pricier ads as a budgeting problem and reaching for the spend slider. You trim the budget and feel responsible. But your cost per customer has not moved, revenue dropped, and your fixed costs now spread across fewer customers. The dashboard reads like control. The month's profit reads worse.

The second break is tired ads, which hide inside the price line. Run the same ads too long and the same people see them again and again. Fewer click. Meta spends more to find anyone who still responds, and that shows up as a rising price even though the auction did nothing (more in our breakdown of where Advantage+ leaks budget). The third break sits after the click: you pay a premium to win a lead, then let the inquiry sit in an inbox for six hours. The most expensive waste in advertising is a lead you paid extra to capture and then ignored.

What to change this week

Stop defending the price per view and defend the payback, meaning how fast a new customer earns back what you spent to win them. Ship more ad variations, faster, so Meta has more to choose between and your cost per sale falls even as the price per view rises. Strengthen the offer before you touch the budget: the share of visitors who buy is the cheapest lever you own, and moving margin through packaging costs nothing per view. Tighten the path after the click with a faster page and an instant reply the moment someone submits your form. Tell Meta to stop showing ads to people who already bought, so you stop paying holiday prices to reach customers you already own. And work on keeping customers longer, the long lever, because a customer who stays two years justifies a much higher price to win. The follow-up sequences that build retention are where that compounds.

How to measure it

Watch the right number, not the scary one. Track what a new customer actually costs you, not what a thousand views cost. If views got pricier but your cost per customer held flat, your ads and offer absorbed the increase. If your ad price is outrunning the market benchmark, the cause is usually tired ads, and the fix is new ones. The metrics worth tracking, and the ones that lie, live in our numbers coverage.

What the work needs

Metathe auction sets the price. You set the ads, the offer, and who gets excluded

GoHighLevelinstant follow-up so expensive leads never go cold

What to watch

Model the holiday jump now: prices climb all year, then step up hard in November and December. Decide in advance which campaigns you will keep funding at holiday prices and which you will pause. Deciding under pressure in November costs more. Watch how often the same person sees your ad (Meta reports this as "frequency"): a climbing number is the early warning that your ads, not the auction, are the bottleneck. And watch whether customers stick around long enough to justify the new prices, because the durable answer to a permanently pricier auction is a customer who is worth more. More under our paid ads coverage, and the full article archive has the rest.