People will tell you anything above 3 is bad. Others run profitably at 6. Neither rule is real. There's no magic number, and chasing one gets founders to kill campaigns that are making them money. Here's how to actually read frequency.
What frequency actually measures
Start with the definition, because most of the panic comes from reading the wrong meaning into it. Frequency is the average number of times your ad was shown to each person you reached. It's an average, which already hides a lot: a frequency of 4 might mean four views for everyone, or two views for half your buyers and six for the other half. The people buying often sit in the high-frequency tail, by the way. Conversions cluster on the third, fourth, fifth exposure, while the average climbs and observers panic.
Why the average matters less than it sounds: a high average on a large pool means repeated exposure, which is how anything accumulates real value, through repetition rather than a single viewing. A lower average on a tiny pool means everyone has seen it plenty. Same number, opposite messages. That's the trap the "keep it under 3" rule walks straight into.
There's no universal damage line
Frequency of 3 can be fatal on one account and 6 can be fine on another. It depends on your audience size, your creative, your market, and whether people still convert after seeing it repeatedly. A small remarketing audience might hit fatigue at 3 because it's a tiny pool. A big broad campaign might run at 6 and keep converting because there's always fresh blood seeing it.
Think about what frequency of 3 means on each. On a 5,000-person custom audience, a frequency of 3 means nearly everyone saw it several times in the window. The pool is genuinely saturated and delivery has nowhere new to go. On a 10-million broad pool at frequency 3, most people touched that week saw it once. The system is just repeating itself inside a small active slice, and there are millions of untouched people behind them. Identical number, one account is at the end of its road, the other is nowhere near it.
So the "keep it under 3" rule is a useful starting guess and a terrible law.
What actually signals damage
Frequency itself isn't damage. Damage is frequency climbing while your click-through rate and conversion rate fall together. That's the combination that says people have seen it too much and stopped responding.
If your frequency is 5 and conversions are steady, there's no damage. Don't touch it. If your frequency is 2.5 and your CTR is sliding, something is wrong already, and waiting for a "safe" frequency won't fix it.
The P&L is the final check, and it overrules the instruments when they disagree. A campaign at frequency 6 that's above break-even is making money. A campaign at frequency 2 that's below break-even is losing money anyway. Frequency warns first, CTR confirms or clears it, and the P&L decides what happens next.
Where the real thresholds sit
Across most e-commerce accounts, the range where fatigue tends to bite is somewhere between 3 and 8, with the exact point depending on how fast your creative refreshes. Brands refreshing weekly often never let frequency climb past 4 or 5. Brands with one hero ad left running for a month might hit 10 plus and see the drop.
Why the range is so wide makes sense once you see the moving part behind it: creative counterweight. A brand that ships two genuinely new concepts a week refills the audience's appetite before it empties, so frequency gets interrupted by fresh messages. A brand running one hero ad spends down the same pool with the same message until response dies, then discovers the number. The refresh rate sets where on the 3-to-8 range your account sits.
Watch the trend, not the level. A frequency rising fast toward 5 is a warning. A frequency parked at 5 for weeks with stable results isn't.
The practical answer
Set a mental watch zone around 3 to 5. When frequency enters it, check your CTR and conversion rate. If they're holding, leave the campaign alone and keep making money. If they're falling, that's your cue to have fresh creative ready.
Don't chase a low frequency number. Chase profit. A campaign at frequency 6 that's still above break-even is worth running. A campaign at frequency 2 that's below break-even is worth killing. The frequency is just one input, the P&L is the verdict.
A worked read, with rounded numbers
Take a typical fashion brand account, numbers rounded so the shape is clear. Broad campaign, Rs.8,000 a day. Frequency had drifted from 2.1 to 4.6 over three weeks, CTR fell from 1.9% to 1.1%, conversions fell with it. The team's plan was to add three interest audiences "to find fresh people." My read was the opposite: frequency 4.6 on a pool of millions, CTR falling in a clean slope, that's the fatigue signature, not saturation. Narrowing would have shrunk the reachable pool and raised frequency further, likely to 6 within days.
We shipped two new concepts inside the week, frequency eased back to 3 as the fresh creative caught, CTR recovered to 1.7%, and ROAS came back to its old range without touching the audience once. Total cost of the fix was creative, which they'd have needed anyway.
The fix there is scope, widen it and stop leaning on so few people. Same symptom, frequency high, opposite disease, and pool size is the number that tells them apart.
I'm Hirdesh Matai, founder of HDM GRO-ads, the science of e-commerce advertising.