State of Paid Acquisition: Mid-2024 Meta Ads Costs and What Comes Next

Ad impressions, up 10%. Average price per ad, up 10%. Revenue, up 22%.

Those three numbers came out of Meta on July 31, and together they are the whole story of paid acquisition in the middle of 2024. There is more ad inventory than a year ago. Each ad, on average, sells for more. And the company that runs the auction keeps making more of the decisions for you.

This is the first of a quarterly read I am calling State of Paid Acquisition. I have run Meta ads for about 10 years, and I read earnings releases the way a farmer reads weather reports: not to predict next week, but to decide what to plant. Here is what the Q2 numbers say, what they do not say, and how I would plan the next quarter if I were spending a few hundred dollars a day.

What did Meta's Q2 2024 numbers actually say?

The headline facts come straight from the release.¹

Advertising revenue was $38.33 billion, up 22%, or 23% excluding foreign exchange effects. Ad impressions grew 10% year over year, and the average price per ad grew 10%. Family daily active people, which counts people using at least one Meta app, averaged 3.27 billion in June, up 7%.¹

The part people skip: headcount

Meta ended the quarter with 70,799 employees, down 1% from a year earlier.¹ Revenue is up 22% with slightly fewer people. I will not pretend one number proves a trend, but it fits what I see in Ads Manager: the platform keeps getting better at running auctions without humans, and it expects advertisers to supply the inputs, which are creative, offers and conversion data.

The wider market grew too

The IAB and PwC report put US internet advertising revenue at a record $225 billion in 2023, up 7.3% year over year.² The same report describes social media advertising growth as having rebounded, which matches the picture Meta's own quarter paints. Meta is not growing in a vacuum. The category it sells in is growing, and a bigger pool of advertising dollars is a bigger pool of competition for the same attention.

Why are volume and price rising at the same time?

Because demand is growing faster than supply, and supply is growing too.

When impressions rise 10% and price per ad rises 10% in the same quarter, advertisers collectively had more places to put ads and still bid up the price of the average one. Meta's release does not explain the reasons, so I will mark this next part as my read and not a fact: a larger pool of advertisers chasing the best placements will push average prices up even as the platform adds inventory.

What price per ad is not

It is not your CPM. It is a blended average across every advertiser, placement and country on the platform. A coach selling a high-ticket program to US buyers and a mobile game buying installs in another region are inside the same number.

What does more automation mean for a small advertiser?

It means the levers you used to pull are being replaced by inputs you feed in.

Meta's own page describes Advantage+ audience as taking your usual audience choices, such as a customer list, gender or an interest, and expanding beyond them, with strict limits you can set for things like minimum age, location, language and custom audience exclusions.⁷ In other words, you suggest, and the system searches. The more reliable your suggestion and your conversion data, the better it searches.

The learning phase rules still apply

Meta's help center says ad sets usually exit the learning phase after about 50 results in the week after the last significant edit, and that editing during the learning phase resets learning. It also recommends realistic budgets, fewer ad sets and avoiding frequent budget changes.⁶

For a small budget, that is the real constraint. If your daily spend cannot produce about 50 results of your chosen event a week, the system stays unstable and your cost per result runs higher. This is why I would rather a small advertiser optimize for a deeper event they can afford than a cheap event that is easy to hit but means little.

Dara Denney gave a version of this for creative tests in a video posted on February 21, 2024: "You don't want to unintentionally tank the performance of an adset or a campaign that's working well."⁸ The same instinct applies to the edits that reset learning: protect what is working while you test.

Signal is the new targeting

Apple's App Tracking Transparency, which became required with iOS 14.5 in April 2021, means apps need a user's permission to track them across other companies' apps and websites for advertising or measurement.³ Since then, browser-only tracking has under-reported what happens after a click. Google's July 22 update added a twist. Chrome will keep third-party cookies and introduce a choice experience for users instead of removing them.⁴ That is a reprieve for browser tracking, not a plan. First-party data, sent from your own systems, is the durable input.

How do you build a break-even number you can trust?

Everything in this piece comes back to one number, so it is worth building it properly. Your break-even cost per acquisition is what you can pay to win a customer before you lose money on them. The formula has three inputs and no tricks:

Collected revenue per customer, minus what it costs you to deliver, minus the profit you need to keep, is the most you can pay to acquire that customer. Everything else in the account is a way of getting under that line.

Stress-test it at plus 10%

Meta's average price per ad rose 10% in a year.¹ You do not have to assume your costs will do the same, but you should know what happens if they do. Take your current cost per acquisition, add 10%, and see whether you are still under break-even. Then add 20% and look again. If the first test breaks you, the problem is not the platform. The problem is that your margin was thin before prices moved.

Count the leak between click and cash

A cheap lead that never shows up is more expensive than a pricey lead who buys. Break-even should be calculated at the paying-customer level, which means your show rate, close rate and payment collection all sit inside the number. This is why I keep saying the ad account is only the first part of the whole revenue system, never the whole thing.

What does an election-year Q4 mean for ad costs?

Probably more competition, and I would plan for it without guessing at a number.

AdImpact now projects a record $10.7 billion in political ad spending for the 2024 cycle, up 19% from 2020.⁵ That figure covers all media, and I cannot tell you what share will land in Meta's auctions. What I can tell you is how auctions behave: when more bidders with big budgets enter, the average price of the same attention goes up, and small advertisers feel it first because they have less room to absorb it.

What should you not do when costs rise?

Rising prices make people do expensive things. I see the same four mistakes every cycle.

Chase the cheapest lead

When costs go up, the temptation is to switch to whatever event is cheapest to hit: a page view, a click, a low-friction form. The dashboard improves and the bank account does not. A cheap event the platform can find easily is also an event that tells it very little about who buys.

Touch the budget every day

Meta's own guidance says frequent budget changes can send an ad set back into the learning phase.⁶ Small accounts are the most tempted to micromanage and the least able to afford it. Set a budget you can hold for a week and judge the week.

Judge a quarter by a bad Tuesday

Auction prices move day to day. A scary Tuesday is data, not a verdict. Look at the seven-day and thirty-day view of cost per paying customer before you cut anything.

Put the whole plan on one platform

The single most useful habit I have built is never letting one platform decide whether my business works. Your email list, your referral flow and your own site do not get more expensive when somebody else's auction does.

What should you do before Q4?

Do these in order. They are cheap, and they compound.

If you want a working rule for the rest of this year, here it is: spend the next two months building the things that make every dollar work harder, and avoid betting the quarter on a cheaper auction. I will come back to this series each quarter and tell you where my read was wrong. If you would rather have a second opinion on your own account than read earnings releases, you can start at ekaistone.com.

Keep building.Ekai