Creative and CPM are connected, and the connection is not a quality score somebody assigns you. Platforms earn more from adverts people respond to, so a creative that earns responses competes more effectively for the same impressions and ends up costing less to deliver.
That is an auction outcome rather than a reward. Understanding which it is changes what you do about it.
This is for you if your cost per thousand is climbing and somebody has told you to make better adverts.
The mechanism behind creative and CPM
Platforms are choosing which advert to show in a slot, and they choose on expected revenue.
Expected revenue is your bid multiplied by how likely people are to engage. An advert that reliably earns responses raises the second part of that, so it can win the impression against a higher bid from somebody duller.
From your side, that surfaces as a lower cost per thousand. You did not get a discount. You won the auction more efficiently.
The distinction matters because it tells you where the effect comes from. Nobody is grading your creative. The system is predicting behaviour, and your advert is either making that prediction favourable or it is not.
What a normal cost per thousand looks like
Triple Whale reported Meta averages across roughly 35,000 brands for the full year 2025: cost per thousand of $14.19, up about 20% year on year, with click-through rate at 2.19% and cost per acquisition at $38.19.
Treat that as directional. Triple Whale is a vendor publishing figures from its own customer base, which is a different thing from an independent measurement.
It is still useful as a rough sense of scale, and the year-on-year direction is corroborated elsewhere. Meta’s own filings show average price per ad up 12% year on year across consecutive quarters, which is a filing rather than marketing and therefore stronger evidence.
So if your cost per thousand rose this year, market pressure is the first explanation to consider, before anything about your adverts.
The trap in reading creative and CPM as a target
Cheap impressions delivered to people who cannot buy are the most expensive thing in an account.
I have a clear example. On a US flag and banner ecommerce account, one ad set targeted a neighbouring state. It produced the highest reach in the account at 17,086 and the most link clicks at 542.
By every efficiency measure on the impression side it looked excellent. It returned 0.25 on ad spend, three purchases at roughly $341 each, about $1,027 spent before it was dropped. The only genuine loser in the account.
The client manufactures locally and delivers same day in its home state. People in the neighbouring state were seeing an offer that was not really available to them. Interest was real, fulfilment was not.
Cheap reach found more of exactly the wrong people, faster. That is the sharpest illustration I have of why creative and CPM together cannot tell you whether an advert is working.
Why chasing the number makes it worse
There is a reliable failure pattern here, and it follows directly from how you would lower the figure.
To reduce cost per thousand you broaden the audience, accept more placements and reduce your bid. Each of those does lower it.
Each also worsens who you reach and where you appear. So the metric improves while the account gets worse, and the improvement is visible weekly while the deterioration shows up a month later in cost per result.
That is why cost per thousand belongs firmly in the diagnostic tier rather than the target tier. It explains why cost per result moved. It should never be the thing you are moving.
How much of the link between creative and CPM you control
Less than the conversation implies, and knowing the split prevents wasted effort.
Seasonality moves it substantially, with Q4 the obvious case. Competitor budgets move it. Audience size moves it, since a small audience is expensive to reach repeatedly. Placement mix moves it, because some placements simply cost more.
Creative moves it too, through the response mechanism above.
So when the figure rises, work through the others before concluding your adverts are tired. A rise that coincides with a seasonal peak, a competitor launch or an audience you narrowed is probably not a creative problem, and replacing creative will not fix it.
The check is whether cost per result moved with it. If cost per thousand rose and cost per result held, the auction got more expensive and your adverts absorbed it, which is the system working.
Where the same logic appears on search
The mechanism rhymes even though the vocabulary is different.
On Google, expected click-through rate feeds into ad rank, so an advert people click more can win a better position at a lower cost per click. Same principle, different metric name.
That is worth knowing because it means the creative work transfers conceptually across platforms even when the assets do not. Better adverts cost less to deliver everywhere, through mechanisms that differ in detail and agree in direction.
It also means the diagnostic discipline transfers. Cost per click on search deserves exactly the same treatment as cost per thousand on social: useful for explaining, dangerous as a target.
What creative and CPM cannot tell you together
Neither number knows anything about who saw the advert.
That is the gap the flag account example sits in. Reach was excellent, cost per thousand was fine, click-through rate was strong. Every impression-side metric agreed, and revenue disagreed with all of them.
So no combination of these figures substitutes for the outcome measure. They describe how efficiently you bought attention, not whether the attention was worth buying.
Keep them for diagnosis and keep the decision on cost per result. That hierarchy is the single most useful thing in this whole cluster, and this pair of metrics is where it gets violated most often, because they are the ones that move first and look most like performance.
What to change this week
Three steps.
Put cost per thousand and cost per result side by side for the last quarter, so you can see whether they moved together or apart.
Then check whether any ad set with unusually cheap reach is also producing unusually poor results, since that combination is the signature of the trap described above.
Then stop reporting cost per thousand as a headline number, and move it into the diagnostic section where it belongs.
Ad creative metrics covers the denominators behind all of these, fund the ad set and read what the algorithm chose is the position underneath, creative drives sales covers what creative genuinely moves, and why cost per lead is rising covers the market pressure half. Organic visibility has the same diagnostic-versus-target confusion, which my SEO work treats the same way. To work out whether your rising costs are the market or the account, book a teardown.