Spend per new creative: the ratio behind a flat month
The metric almost nobody tracks, and the one that explains more flat months than any targeting change.
If spend is climbing and the number of new creatives you launch each month is flat, the account will quietly get worse and it will look like the algorithm turned on you. It didn't. You ran out of things to show people. The measure is spend per new creative, month by month: total spend divided by how many distinct creatives first delivered that month. You are not looking for a target number — there isn't one — you are looking at whether your own line is rising. When it rises, performance decays on a delay of weeks, which is exactly long enough to blame something else.
This is the metric almost nobody tracks and it explains more flat months than any targeting change ever will.
The pattern is always the same. Something works, so you put more budget behind it. More budget means more people see the same creative more often. The audience that responds to that particular ad gets used up — a different problem from the idea itself wearing out, and the two need opposite fixes. Performance sags. You look at targeting, bidding, placements, the pixel — and the actual cause is that you have been spending more against a fixed amount of creative.
The measure
| What you look at | What it tells you |
|---|---|
| Spend per month | Nothing on its own |
| New creatives launched per month | Nothing on its own |
| Spend ÷ new creatives, per month | Whether you are asking more and more of the same work |
| The direction of that line over six months | Whether a flat month was predictable |
Ask for it like this, with your own window:
By month for the last six months:
total spend, and the count of DISTINCT creatives that first
delivered in that month. De-duplicate by creative_id — the same
creative reused in a new ad set is not a new creative.
Then show spend divided by new creatives, per month.
Echo the exact months back before the numbers.This is the same de-duplication problem that makes
ROAS sorting misleading, and it distorts
this measure in the flattering direction — you look like you are producing more than you are. If
you take one thing from this page, make it that creative_id is the unit, not the ad.
Reading the line
-
Rising steadily THE COMMON CASE
Spend growing, creative output flat. You are heading for a soft month and the cause will not be visible in the metrics you normally check. The trade: the fix is production, which is slower than any setting you could change — which is exactly why it gets skipped.
-
Flat
Output scaling with spend. This is what you want and it is unglamorous. The trade: it says nothing about whether the creative is any good — volume is necessary, not sufficient.
-
Falling
Producing faster than you are spending. Fine while you are searching for an angle, expensive as a permanent state. The trade: if nothing is winning, more volume is not the answer — go back to what the winners had in common.
-
Spiky
Batches then nothing. The gaps are where fatigue accumulates. The trade: a steady trickle beats a quarterly burst, even at the same total.
Fatigue is not really a creative problem. It is an arithmetic one: budget went up, supply didn't.
Keeping this line flat is the whole point of running a creative refresh every week rather than when things look bad. See how the weekly refresh works →
What to do when it is rising
Do not start from a blank page. Take the creatives that already earned their spend, write down what each one actually is — the hook, the visual world, the offer, the angle — and make variants that keep the world and change the scene. That is a much shorter job than inventing, and it is the one an AI is genuinely useful for: it can read the winners back to you, including the images themselves, and draft the next batch paused for you to pick from.
- Counting ads instead of creatives. Inflates the number in the direction that makes you feel fine.
- Waiting for the dip to act. The decay lags the cause by weeks, so by the time it shows, you are already behind.
- Confusing new ad sets with new creative. New audiences against old creative buys you a little time and no supply.
- Treating the ratio as a benchmark. There is no right number. Your own trend is the whole signal.
Try this tonight Run the prompt above for six months. If the line is rising, you have just explained a flat month without touching a single setting.
This is question three of the four-question audit. If you have not connected an account yet, setup takes about three minutes.
Frequently asked questions
What is a good spend-per-new-creative number?
There isn't one, and anyone quoting a benchmark for your account is guessing — it depends on price point, margin and how fast your audience saturates. The direction of your own line is the signal.
Do small variants count as new creatives?
A genuinely different scene or hook counts. A headline swap on the same image is a test, not supply. Be strict about this or the measure flatters you.
Isn't this just creative fatigue?
Fatigue is the symptom people notice. This is the arithmetic underneath it, and unlike fatigue it is visible before performance drops rather than after.
How many new creatives per month should I aim for?
Enough to keep the line flat as spend grows — which makes it a function of your spend trajectory, not a fixed quota.
Keeping the line flat, every week
The Claude Ads Operator is the routine that makes creative refresh a weekly habit rather than a panic — winners read back, variants drafted paused, your call on Friday. No ROAS promises anywhere in it.
See what's inside — $67 founding pricePre-sale: the modules are still being recorded and founding members get the checkout link first. No charge today. Public price $97.