What "frequency" actually measures
Frequency is a ratio, not a count. It is impressions divided by reach — the average number of times each person in your audience saw the ad over a chosen window. A frequency of 3.0 over seven days means the average person saw your creative three times that week.
That "average" hides a lot. If frequency reads 3.0, some people saw the ad once and some saw it a dozen times. This is why a single threshold can never be a hard rule — the distribution behind the average matters as much as the average itself.
The window matters too. Frequency measured over a lifetime will always look higher than the same ad measured over seven days. When you compare against any benchmark, compare on the same window. Most fatigue thresholds are quoted on a rolling seven-day basis.
The frequency thresholds practitioners actually use
Meta does not publish an official "yellow at this number, red at that number" label. The thresholds below are practitioner conventions the industry uses to read frequency before the cost penalty shows up in the delivery column. Treat them as prompts to look, not automatic kill triggers.
For cold prospecting, GoodMorning puts the early-warning zone around a 2.5 frequency and the "refresh now" zone around 3.5, where the ad is usually starting to lose money. A separate benchmark roundup from AdAmigo lands in the same range: keep cold frequency under 3.0, treat 2.5 as a warning and 3.5 as critical.
Warm audiences behave differently. The same AdAmigo benchmarks put the retargeting sweet spot around 4.0 to 6.0, with fatigue arriving later because those people already know you. If you apply a cold-audience threshold of 3.5 to a retargeting ad set, you will kill winners early.
Vertical and price point move the line as well. A low-priced impulse product can burn out faster than a considered, high-ticket purchase where people need several exposures before they buy. The number is a starting hypothesis, not a verdict.
Why cost per result is the number that actually matters
Here is the trap in reading frequency by itself. Frequency going up is not inherently bad — a healthy, well-targeted ad can run at 4.0 and stay profitable. Frequency going up while cost per result also rises is the fatigue signal you can trust.
The reason is auction mechanics. As an ad wears out, click-through rate falls, which lowers your estimated action rate, which raises what Meta charges you to keep showing it. GoodMorning notes cost per result roughly doubles by the time Meta's own creative-fatigue flag fires, which is why the earlier 2.5 and 3.5 frequency reads give you a head start on the platform's own warning.
The decay compounds as exposures pile up. AdAmigo reports that once the same person has seen an ad five or more times, costs can rise roughly 50 to 80 percent while click-through rate drops 40 to 55 percent. Those are aggregate practitioner figures, not a guarantee for your account, but the direction is consistent: attention erodes, and Meta prices that erosion back to you.
So the practical rule is a pairing, not a threshold: plot frequency and cost per result on the same timeline. When both rise together on a single creative, that creative is fatiguing. When cost rises but frequency is flat, look elsewhere — a broken pixel, a more expensive auction, or a landing-page change.
Reading fatigue against profit, not just ROAS
Every article on this topic stops at "refresh your creative." The part they skip is the one that decides whether fatigue is actually costing you money: what cost per result can you afford in the first place?
That ceiling comes from your margin, and it is pure arithmetic. Break-even return on ad spend equals one divided by your contribution margin — the share of revenue left after cost of goods, shipping, and fees, before ad spend. Say you sell a print-on-demand hoodie: revenue of $50, contribution margin of 50 percent, so $25 of gross profit per order. Break-even ROAS is 1 ÷ 0.50 = 2.0x, meaning you can pay up to $25 to acquire an order before you lose money.
Now watch what fatigue does to that. Say a fatiguing creative pushes your cost per purchase up as frequency climbs. Walk the arithmetic: if cost per purchase moves from $15 to $22 while your break-even sits at $25, your per-order profit shrinks from $50 × 0.50 − $15 = $10 down to $50 × 0.50 − $22 = $3. The ROAS headline may still read above 2.0x, but you have lost 70 percent of your profit on those orders.
This is also why average ROAS lies during scaling. A campaign averaging 4.0x can have a marginal ROAS near break-even on its most recent budget, because each new dollar reaches a less-responsive slice of the audience. Fatigue and diminishing returns stack: the ad gets more expensive per result at the same time each extra dollar buys a worse customer. If you want the full model for scaling on the marginal number instead of the average, we walk through it in our guide to profitable ad scaling.
What to do when the threshold trips
First, confirm it is really fatigue and not measurement. Reconcile Meta's reported revenue against your actual store revenue for the same window. If the backend is steady but Meta shows decay, the problem may be a dropped pixel or a changed attribution window, not a tired ad.
If it is genuinely fatigue, the highest-leverage fix is new creative, not a new audience. Since Meta's Andromeda retrieval rebuild, the creative itself is the primary targeting signal — the hook and format decide who sees the ad more than manual interest lists do. Refreshing the concept both resets attention and re-points delivery. A steady cadence of a few fresh concepts per week keeps a new winner ready before the current one burns out.
Be careful how you make the change. Swapping creative inside an existing ad set can count as a significant edit and re-trigger the learning phase, which needs roughly 50 optimization events within a seven-day window to exit, per Jon Loomer's read of Meta's guidance. If you fragment budget across too many new ad sets, each one pays that learning tax separately. Consolidate where you can.
If frequency is climbing because the audience is simply too small for the budget, widening the audience or leaning on broad and Advantage+ delivery slows the frequency creep. How you structure that budget — one campaign budget versus per-ad-set budgets — changes how fast fatigue spreads; our breakdown of CBO versus ABO campaign structure covers when each contains fatigue better.
Where a profit view changes the decision
Notice that every step above needs one number the ad platform never shows you: true per-order profit. Meta reports ROAS and cost per result. It does not know your cost of goods, your Printify or Printful print cost, your Stripe fees, or your shipping. So the moment a fatiguing ad crosses from "expensive" to "unprofitable" is invisible inside Ads Manager.
This is the gap PodVector is built to close. It connects Shopify, Meta Ads, Google Ads, Printify, and Printful, and computes true per-order profit — so a rising cost per result can be read against the actual margin it is eating, not a ROAS number that ignores your costs. Victor, its AI operator, analyzes that combined data and proposes moves; with your approval he acts on the Shopify side. He reads your ad data but does not touch your ad account, and he is not a dashboard — the point is a decision, not another chart.
One profit lever worth pairing with any fatigue fix: raising average order value lowers the break-even ROAS your ads must clear, which buys you room even as a creative ages. Post-purchase upsells add margin at zero extra acquisition cost — see our rundown of post-purchase upsell tools and the broader set of Shopify apps to increase AOV.
FAQs
What is a good frequency for Meta ads?
There is no universal target. For cold prospecting, practitioners generally keep seven-day frequency under about 3.0 and start watching around 2.5, according to AdAmigo. Retargeting audiences run comfortably higher, often in the 4.0 to 6.0 range. The "right" number depends on your creative volume, audience size, and how considered the purchase is.
Does high frequency always mean my ad is fatiguing?
No. Frequency on its own is not a kill signal. A well-targeted ad can run at 4.0 and stay profitable. The trustworthy fatigue signal is frequency and cost per result rising together on the same creative. If frequency climbs but cost per result holds steady, the ad is not fatiguing yet.
At what frequency should I turn off a Meta ad?
Do not turn it off on a frequency number alone. Turn it off when cost per result rises past what your margin can afford. That ceiling is your break-even ROAS, which equals one divided by your contribution margin. A 50 percent margin sets break-even at 2.0x; once fatigue drives cost per acquisition past your allowable customer acquisition cost, the ad is losing money regardless of the frequency reading.
Why did my cost per result rise but frequency stayed flat?
That pattern usually points away from creative fatigue. Likely causes are a broken or under-reporting pixel, a more expensive auction (seasonality or new competitors bidding), a landing-page or checkout change hurting conversion, or a recent significant edit that reset the learning phase. Rule those out before blaming the creative.
How often should I refresh creative to stay ahead of fatigue?
Enough that a fresh winner is ready before the current one burns out — a function of your spend and audience size, not a fixed number. A common practitioner cadence is a few new concepts per week for actively scaling accounts. Smaller accounts should test fewer creatives for longer so each one gathers enough purchase data to read cleanly.
Does refreshing creative restart the learning phase?
It can. Changing creative inside an existing ad set may count as a significant edit and re-trigger learning, which needs roughly 50 optimization events in a seven-day window to exit, per Jon Loomer. To limit that tax, avoid spreading budget across many tiny ad sets, and make changes deliberately rather than constantly.