Meta CPMs Are Up. What That Means for Pop, Push and Native Buyers in Q4
Northbeam, which tracks more than a thousand ad accounts, reported on October 22 that average Meta CPMs were already up about 25%, before the usual step-up around November 1. Meta itself said this month that ad sets getting fewer than roughly 50 conversions a week cost about 28% more per purchase. If you buy on Meta, that's your problem for the next ten weeks.
If you buy pop, push and native on ad networks, it's partly your opportunity. Partly.
When social gets expensive, two things happen on our side. Some advertisers who normally live inside Meta go looking for cheaper volume and test PropellerAds, Adsterra, RichAds or Zeropark for the first time. And affiliate offers that only just worked on social stop working there, so the affiliates running them bring them over too. More advertisers in the same pop and push auctions means bids creep up on the good sources. The cheap sources stay cheap, usually for a reason.
So I don't buy the idea that ad networks are a bargain in Q4 by default. They can be. You have to earn it at the source level.
The first thing I'd do right now is clean up whitelists. Pull the last 30 days by source or zone ID, look at cost per conversion and post-click behaviour, and be honest about which sources are carrying the campaign. On most pop campaigns a small group of sources does the real work and a long tail eats budget. In a quiet month you can tolerate that tail while you test. In Q4, with newcomers pushing bids on the same inventory, the tail gets more expensive and it doesn't get better. Cut it, move the winners into their own campaigns, and bid on them deliberately.
Then check traffic quality before you scale anything. More money flowing into a format attracts more junk. I'd look at time on page, scroll depth and the ratio of landers to offer clicks per source, and I'd use whatever bot filtering the network offers plus your own tracker rules. A source with a great CPA on day one and zero engagement on the lander is a source you'll regret by day five. Kill rules should be written down before launch. Spend a fixed multiple of your payout on a source with no conversion and it's out. No arguing with it at midnight.
Offers matter more than people admit. Push and pop users aren't in shopping mode the way a Meta user scrolling past a retargeting ad might be. They clicked a notification or landed in a new tab. Seasonal hooks help in Q4, and simple offers with a short flow convert better than anything that needs a long form. Sweepstakes, utilities and single-step leads usually hold up. If your offer needs trust and a credit card on the first visit, pop traffic will be rough on it, holidays or no holidays.
Tracking is the part that quietly breaks things this time of year. Postbacks fail, a network changes a macro, an advertiser caps an offer on a Friday and nobody tells you. Run your tracker's postback test on every campaign before you raise budgets, and compare network-reported conversions to tracker conversions daily. If the gap is growing, stop and fix it before you optimise on bad data. Pausing your best sources because a postback died is a painful way to lose a December.
Native sits in between. It's closer to social in user intent, and it gets more crowded in Q4 because brand money spills over there. I'd treat native as the place to test angles that worked on Meta, with tighter caps than you'd use on pop, and widgets blocked fast when they don't convert.
Now the part I'd actually plan for. Marketing Brew reported that some buyers are shifting budget into late December and January, the stretch people call Q5, when big advertisers pull back and auction prices tend to ease. On ad networks the same pullback tends to show up. The tourists go back to social once it's cheap again, and the sources they were bidding up cool off.
That's when I want to be testing. The first week of December is too late. Which means the prep happens now:
- whitelists and blacklists exported and up to date
- two or three offers approved with the network or affiliate manager in advance, with caps confirmed for the holidays
- landers built and postbacks tested
Affiliate managers go on holiday too. If you email for an offer approval on December 26, you might get an answer on January 5.
The practical step for this week: export your top sources by zone ID from every network you run, save the list with today's bids and CPAs, and set a calendar check for December 27 to compare the same sources at the same bids. That comparison tells you whether Q5 is cheap on your traffic, and it's the number your January budget should be based on.