Hey,

Meta reported earnings last week. The coverage was all about AI spending and data centers, and buried under it was the one line that actually lands on your P&L:

The average price of an ad rose 12% in a single year.

That's across every advertiser in the auction, good and bad. Same audiences, and every unit of attention now costs 12% more than it did last summer.

Impressions grew 14% too, which means Meta created more inventory and prices still climbed. More advertisers are bidding on the same people every quarter.

Here's the part most founders get wrong: they treat this like weather.

"CPMs are brutal right now, we'll scale when they calm down."

I've heard a version of that sentence every year I've done this work, and every year the baseline resets higher.

Good months and bad months, sure. The ten-year direction is up.

You can't wait this one out.

The only real move is making your economics immune to the tax, and there are three ways to do it. The best brands we work with run all three.

Before the routes, one uncomfortable implication.

If your ROAS target was set a year or two ago, it's stale. A target that gave you real margin at last summer's prices might leave you at breakeven today, and you'd never notice, because the target looks the same while the costs underneath it moved.

This is how brands "hit their numbers" all the way into a cash crunch.

 

Route 1: Make every click worth more

You can't control the price of attention. You fully control what a visitor is worth once you've paid for them.

A hobby brand I work with hit a wall at an $11 cost per new customer against a $46 first order.

Squeezing CAC lower stopped working, so we went the other way and engineered the first order toward $70 with bundles, an above-AOV gift threshold, and a guarantee-backed starter kit.

Same products, same traffic, and suddenly there was $16 to $20 of acquisition headroom per customer instead of $8.

That single move absorbs years of 12% attention inflation. When your first order affords a higher CAC, rising CPMs stop being scary and start being a line you plan around.

 

Route 2: Earn cheaper attention instead of buying it retail

The auction has a discount nobody talks about: relevance.

Meta charges engaging ads less per impression, because engaging ads keep people on the platform. Fresh creative at real volume is how you qualify for that discount.

The brands paying the least per unit of attention are running 3 to 5 actually new concepts every week, killing losers inside days, and repackaging winners into new formats before fatigue sets in.

The brands paying retail are riding one tired winner at frequency 4 and wondering why their CPMs are "unlucky."

Creative volume was already the biggest lever in paid media. Every price increase makes it bigger.

You don't need a content studio for this. You need a system: scripted hooks with room for creators to be human, winners multiplied into split-screen and voiceover versions instead of retired, and a testing budget that never touches your scaling campaigns.

A health brand we work with runs that machine every week at $300K a month in spend, and that system is why the spend works.

 

Route 3: Stop renting all of your attention

Every subscriber on your email list is attention you bought once and get to reuse for free.

No auction, no 12% tax, no algorithm deciding your reach.

For a brand with real repeat purchase, email should carry 25 to 30% of revenue. Most brands I audit sit near half that, which means they keep re-buying attention they already paid for.

The gap is almost always flows, the automations that sell while you sleep. A welcome series, abandoned checkout, and post-purchase flow that actually reflect your current catalog will move more revenue than most founders' entire ad-testing budget, and they only need to be built once.

Every point of revenue you shift to owned channels is a point that Meta's pricing team can never touch again.

It's also the strongest Q4 argument there is: the list you build in August converts in November for free.

 

What NOT to do

Two reactions I watch founders have to rising CPMs, both expensive.

Don't panic-cut spend the moment efficiency dips, especially mid-test. You'll burn the learning you already paid for and restart from zero next month at even higher prices.

And don't chase cheap CPMs into junk placements and ultra-broad audiences that "lower costs" while lowering everything else with them.

You're shopping for customers, not impressions, and the cheapest impressions usually bring the most expensive customers.

If the whole auction moved 12%, that's physics, and firing your media buyer over physics fixes nothing.

 

The stress test to run this week

Take your current numbers and re-run your breakeven with CAC 12% higher, the worst realistic case where the full price rise lands on you.

That's it.

Fifteen minutes, and our free calculator does the math properly, including the costs founders forget.

If the math still works, you have a buffer, and you can plan Q4 from strength.

If the math breaks, you just learned that your unit economics only worked at last year's prices, and you found out in August instead of during Black Friday week, when seasonal CPM spikes stack on top of the structural rise and the lesson costs 10x more.

One more timing note.

Attention only gets more expensive from here through December, which makes this stretch the cheapest attention you'll buy for the rest of 2026. That has a strategic consequence big enough to get its own email in two weeks.

For now, just know the discount window is open, and it closes a little more every week.

The 12% already happened. What you do between now and November decides whether it hurts.

Talk soon,
Andrej

PS: Want to know your actual spend ceiling before Q4? Reply "ceiling" and I'll send you the Spend Ceiling Worksheet we use with clients. Nine inputs from your P&L, and it tells you your max profitable CAC three different ways.

PPS: Me, handing every founder their 2026 media plan →