CPM goes up. First reaction: panic. Traffic is getting more expensive. The algo is eating the budget. Time to pull back.
I have been on both sides of this — as an in-house optimizer and now at TMG managing campaigns across Baidu, Douyin, Xiaohongshu, and WeChat. And the truth is: a rising CPM tells you almost nothing on its own. You need the full picture.
Here is a real campaign example, three things that actually drive CPM up, and a quick framework to figure out whether you should be worried or excited.
CPM + CTR + CVR = the only trio that matters. If CPM goes up but CTR and CVR also go up, congratulations — you found better audiences. If CTR flatlines while CPM climbs, that is when you have a problem.
📊 The Case That Changed My Mind
Early in a consumer e-commerce campaign — off-platform traffic driving (pulling users from social into an external store) — the first week looked decent:
First glance at the dashboard: CPM jumped from ¥28 to ¥36. That is a 29% increase. If you stopped there, you would call it bad and start cutting budgets.
But look at what else happened. CTR nearly doubled. CVR climbed. CPA dropped from ¥120 to ¥72 — a 40% reduction in cost per acquisition. The higher CPM was not a sign of waste. It was the system figuring out who the real buyers were.
We made two changes: (1) rebuilt the first 3 seconds of the video creative — lower drop-off, more click peaks earlier, and (2) tightened the audience targeting. The algo responded by pushing the ad into a higher-value traffic pool. More expensive per 1,000 impressions, but each impression was now worth way more.
🔍 Three Things That Drive CPM Up
Not all rising CPMs are the same. Here are the three most common scenarios we see in practice:
1. 🏟️ More advertisers, same pool
Classic supply-and-demand. Right before 618, Double 11, or any peak shopping season, everyone floods in. More bids → higher floor prices. Your CPM goes up, but your CTR, CVR, and CPA usually don't improve. That is pure market inflation — and the only real lever is timing or platform diversification.
2. 🎯 You hit a better audience
This is the case study above. The oCPM model learns over time and starts bidding on higher-intent users. Those users cost more to reach — they are being fought over by everyone — but they convert at a higher rate. CPM up, CTR up, CVR up, CPA down. This is the version you want.
3. 🪣 Your audience pool is too small
Over-optimized targeting. You narrowed the audience so much that the system runs out of room to explore. CPM keeps climbing but nothing else moves. No CTR improvement, no CVR lift, no CPA drop — just expensive reach into the same tiny pool. The fix: widen the targeting or kill the campaign.
🧠 The CPM-CTR Decision Matrix
Here is a framework we use internally. Four combinations, four answers:
Better creatives, better targeting. Watch CPA. If CPA holds or drops, scale up.
Paying more for the same (or worse) engagement. Competition spike or creative fatigue. Fix the creative or pause.
Cheap traffic nobody wants. Your ads are landing in front of the wrong people. Check your targeting or refresh the creative.
You found a sweet spot. Low competition + high relevance. Double the budget and ride it.
💡 The Takeaway
CPM is a single data point. Treating it like the only signal is how good campaigns get killed early.
What actually matters: CPM + CTR + CVR → CPA. Look at the chain, not the first link. If CPM goes up and the rest of the chain holds or improves, you are in good shape. If CPM goes up and everything else flatlines, that is your real warning.
We run paid media across Baidu, Douyin, WeChat, Xiaohongshu, Bilibili, and Bing China for international agencies and brands. If you want a second opinion on your campaign data — or just want to talk shop — reach out.