State of Paid Acquisition: Q4 2025 and the Start of 2026

In 2025, Meta stopped growing mainly by charging more per ad and started growing by showing more ads. That one shift should change how you plan 2026.

Quarterly reads like this one are not exciting on purpose. I am not here to predict the stock or to tell you Meta is about to get cheap or expensive. I am here to answer three practical questions after a big year: what happened to price, what happened to volume, and what changed in how we are allowed to read the results.

I have run Meta ads since 2014 and directed ads for three 8-figure brands and one 9-figure brand. The previous read, Q3 2025 and the Measurement Reset, sets up this one. My rule when I read these releases: take the platform's numbers seriously as a weather report, and never mistake them for your own forecast. Weather tells you whether to bring a coat. It does not tell you where your ship is going.

What did Meta report for Q4 2025 and the full year?

Strong numbers, and a clear signal that advertisers kept spending.

The quarter

For the fourth quarter, Meta reported total revenue of $59.893 billion, up 24%, and advertising revenue of $58.137 billion, up 24%. Ad impressions rose 18% year over year and the average price per ad rose 6%.¹

The year

For all of 2025, Meta reported advertising revenue of $196.175 billion, up 22%. Impressions rose 12% and average price per ad rose 9% for the year.¹ Meta's CEO, Mark Zuckerberg, summed it up in the release.

None of that is a surprise. The more useful question is how the growth was built.

Did ads get more expensive in 2025, or just more plentiful?

Both, but the mix shifted, and that matters for your plan.

Here is the path through the year, from Meta's four quarterly releases. The first number is growth in ad impressions, the second is growth in average price per ad, both year over year.⁴³²¹

Look at the shape. In the first quarter, price did most of the work. By the fourth, volume did. Impressions climbed every quarter while price growth cooled in Q4.

What Meta says drove it

On the Q4 call, CFO Susan Li said impression growth was driven primarily by engagement and user growth, with ad load optimizations contributing to a lesser degree. She tied the 6% rise in price to increased advertiser demand, largely driven by improved ad performance.⁵ Read that carefully: Meta's explanation is that better-performing ads attracted more demand, which pushed prices up.

What "average price per ad" is and is not

This figure is a blended number across every advertiser, placement and region on Meta's platforms. It can move because of mix, for example when impressions grow in a cheaper surface or market, as well as because of auction pricing. It is a market temperature reading, not a CPM you will see in your account. I would use it to judge direction, then trust my own data for decisions.

What changed in how Meta finds buyers?

The delivery system now leans on large models, and that changes what feeds it.

Andromeda picks the candidates

Meta's engineering team described Andromeda in December 2024 as the retrieval layer that hands a few thousand ads, out of a vast pool, to the ranking models. It reported a 6% recall improvement for retrieval and an 8% ads quality improvement on selected segments, and said Advantage+ and generative tools are multiplying the number of creatives, which it designed the system to handle.⁷

GEM learns from the whole picture

In November 2025, Meta described GEM, its largest ads foundation model (my plain-English version), trained to transfer what it learns to the models that rank ads in Facebook and Instagram. Meta said it learns from ad content and engagement data across surfaces, and that GEM contributed to a 5% increase in ad conversions on Instagram and a 3% increase on Facebook Feed in the second quarter.⁶

What that means for you

I am careful here: those are Meta's own figures about Meta's own systems, and they describe averages across the platform. What I take from them as an operator is practical. The systems are built to match a very large pool of ads to individual people, so a thin creative bench gives the system little to work with. And the system learns from the outcomes you send it, so the quality of your conversion data matters more than it did when targeting was manual.

What changed in how you read results?

The ruler changed on January 12, 2026. I wrote about the week it happened in View-Through Is Gone.

Meta stopped returning the 7-day view and 28-day view attribution windows from its Ads Insights API. It still returns the 1-, 7- and 28-day click windows plus 1-day engaged view and 1-day view, and it set new history limits: 13 months for unique-count fields and hourly breakdowns, six months for frequency breakdowns, and up to 37 months for total values.⁸ According to trade coverage, asking for a removed window now gets you blank results, not an error message.⁹

Do not compare across the seam

If your Q4 dashboards counted 7-day view conversions and your Q1 dashboards do not, a drop in reported results might be only the ruler changing. Re-baseline on windows that exist on both sides, and annotate January 12 on every chart.

Treat platform incrementality as a signal

Meta also offers an incremental attribution setting. Seer Interactive's test covered six accounts and roughly a million dollars of spend. Meta reported 87% of conversions as incremental on average, versus 67% when measured in GA4, and Seer advised not blindly trusting Meta's numbers and cross-referencing with other tools.¹⁰ That is a sensible stance for 2026: use platform numbers for trend and your own records for verdict.

What does Meta's 2026 outlook tell you?

Meta guided to continued growth and heavy spending.

For Q1 2026, Meta guided to total revenue of $53.5 billion to $56.5 billion. It also said it expects 2026 total expenses of $162 billion to $169 billion and capital expenditures of $115 billion to $135 billion, driven mainly by infrastructure costs.¹ Fourth-quarter operating margin was 41%, down from 48% a year earlier.¹

My read is modest and I want to label it as a read: a company investing at that scale wants its advertisers to keep winning, which pushes it to keep improving ad performance, and improved performance pulls more advertisers into the auction. I would plan for competition for attention to stay high rather than expecting a quiet, cheap year.

What mistakes does a year like this invite?

Three, and all of them are cheap to avoid.

Reading platform growth as proof your ads will work

Meta's revenue rising 22% in a year says that advertisers in aggregate found it worthwhile to spend. It says nothing about your offer, your price or your page. A growing auction can be a bigger opportunity and a more crowded room at the same time.

Judging creative on a window that changed

If a creative looked weaker in January than in December, check the attribution window before you kill it. The ad may have lost view-based credit that was never going to show up again, while its click performance stayed exactly where it was.

Charley Tichenor of Disruptor School took a blunt view of fatigue in a video posted on January 11, 2026: "What most people call ad fatigue is usually just bad campaign management. It's operator error."¹¹ I would not go that far for every account, but his habit is right: judge a test on total revenue minus ad spend before you blame the creative.

Plays here, from YouTube

Adding automation without guardrails

The more of the setup the platform handles, the more it matters that you read what is switched on before you publish. Defaults are decisions someone else made for you. Keep a record of which Advantage+ options are active on each campaign, so you can explain a change in results later.

How would I set a 2026 media plan?

Start from your economics, not from Meta's. The January half of this plan is in Q1 2026 Prep.

What I would avoid

I would not raise budgets because the platform's revenue grew, and I would not cut them because prices rose. Neither tells you anything about your offer. I also would not chase every new feature at once. Change one major variable at a time, so you can tell what caused what.

The short version

2025 was a year when Meta grew by showing more ads to more people, helped by models that matched ads to buyers more efficiently, and ended it by changing how the reporting works. For a small advertiser, the ruler and the creative bench matter more than the weather: know which windows you are using, and keep your supply of ideas ahead of the system's appetite.

Do those two things, set the ceiling from your own margin and keep reading the platform's numbers as weather. That is the logic behind the systems I build for clients. Next quarter, I will do this again with Q1 2026.

Keep building.Ekai