You run weekly ACoS reviews. You know your target. You notice when something shifts. That kind of discipline is what keeps an account healthy. But ACoS creep is designed to slip through even that routine, because it does not announce itself in any single week.
It is the ACoS that goes from 28% to 31% across three weeks, then to 34% over the following four, then to 38% as the trend holds. No single week triggers a red flag. Each incremental change falls within normal fluctuation ranges. But the direction is persistent, and the cumulative effect is that you are spending a third more per dollar of sales than you were five weeks ago, with no obvious moment to point at as the cause.
That is ACoS creep. And it is harder to catch than a spike precisely because each individual day looks fine, which means your usual check-in cadence is necessary but not sufficient on its own. The question is what you compare it against, and what watches for it in between.
What ACoS creep is
ACoS creep is a sustained, gradual increase in advertising cost of sales over a period of weeks, with no single identifiable event as the trigger. The ACoS does not jump. It drifts. Because each incremental change falls within normal fluctuation ranges, it often escapes routine review until the cumulative damage is visible at month-end.
The defining characteristic is that it compounds. A 3-point increase in week one is unremarkable. An additional 3 points in week two feels like variance. By week four, you are 12 points above baseline, and the individual weekly changes are meaningless because the problem is the direction, not any single reading.
The practical consequence: even weekly reviews can miss creep if you are comparing this week's number to last week's rather than to a longer baseline. The direction of the trend across 4 to 8 weeks is what separates creep from noise. A single weekly reading tells you where you are. A series of weekly readings tells you where you are going.
Why it happens
ACoS creep rarely has a single cause. It accumulates from multiple small pressures operating simultaneously. The five most common:
Keyword drift. Over time, Amazon's match type behavior tends to expand the range of search terms your keywords attract. What once matched tightly starts matching more broadly. Each new term individually looks like normal variance. Together, they pull conversion rate down gradually as traffic quality erodes.
Negative keyword gaps. The search term report surfaces irrelevant or low-converting queries every week. Without consistent negative keyword additions, those queries accumulate spend without generating sales. This is one of the most common sources of creep and one of the easiest to address once you name it. A thorough read of your search term report will surface most of it.
Rising competitor bids. If competitors in your category increase their bids, Amazon's auction dynamics push your effective position down unless you respond. The account appears to be running normally. Spend is going out. But the traffic quality has quietly shifted because you are no longer winning the same placements you won before.
Conversion rate erosion. Seasonal shifts, listing changes, or price adjustments can gradually lower your conversion rate. If your bids do not adjust to reflect a lower conversion rate, ACoS rises proportionally. This is particularly common in the back-to-school and post-Prime-Day windows, when buyer intent patterns shift and what worked in June works less well in August.
Budget utilization timing. If your daily budget is running out earlier as the account scales, your campaigns concentrate spend in earlier dayparts where conversion rates may be lower. The account spends the same amount but at worse efficiency as the window narrows. Tracking this with an understanding of how budget pacing works will help you recognize the pattern.
How to detect it
The standard weekly ACoS view will not catch creep reliably, because each week's number looks like a small change. You need a comparison that surfaces the cumulative pattern. Two approaches that work:
The 7-day vs. 28-day rolling comparison. Pull your current 7-day ACoS and your 28-day rolling ACoS in the same view. If the 7-day is consistently higher than the 28-day by more than 8 to 10 percentage points, you are in a creep pattern. The 28-day average is acting as your baseline, and the current week's performance is diverging from it in a sustained way.
The weekly trend chart. Export your weekly ACoS from your campaign manager for the last 8 weeks. You are not looking for any single week. You are looking for slope. A gradual upward slope sustained over 4 or more weeks is the signal. A flat or bouncing line is noise. This view is what separates creep from variance.
Once you have identified a creep pattern, the next step is tracing where it is coming from. Pull the search term report for the same 28-day window and sort by spend. Look for terms consuming budget with zero or one conversion. Look for terms with ACoS significantly above your target. These are almost always the source. The pre-automation audit checklist walks through this diagnostic in full.
What creep reveals about your account
ACoS creep is a signal that the world around your campaigns has shifted faster than your last review captured. Keyword drift is gradual. Competitors adjust bids without announcement. Seasonal conversion rate changes are slow-moving and can be invisible week to week, even to an attentive account manager.
A weekly review catches obvious problems, but sustained directional changes can accumulate between those checkpoints. That is not a failure of attention. It is the nature of gradual drift. The gap between reviews is where creep lives, and filling that gap is what keeps accounts efficient over time.
This is also why creep compounds if not addressed early. Keyword drift widens. Negative keyword gaps grow. Each week the account drifts further from where it was optimized. What counts as a good ACoS depends heavily on your margins, and a crept ACoS that was once acceptable can cross into unprofitable territory without any single visible trigger.
How to stop it
The manual response to catching creep is straightforward, if not quick:
- Pull the 28-day search term report and add negatives for any term with meaningful spend and no conversions
- Review placement data monthly. Top-of-search vs. rest-of-search conversion rate divergence points to bid modifier adjustments
- Compare your current conversion rate by campaign to what it was 30 and 60 days ago. Adjust bids downward where conversion rate has dropped and has not recovered
- Check your daily budget utilization time. If campaigns are exhausting budget 2 to 3 hours earlier than they were 30 days ago, something has changed in your spend pattern that warrants investigation
Done consistently on a weekly cadence, this process keeps ACoS stable. The goal is to make the gap between reviews as small as possible and to add an automated layer that catches threshold breaches on the days in between.
That automated layer is a performance rule that fires when your ACoS crosses a threshold above your target. The rule can be set to send an alert, cut budget, or pause, depending on how high the ACoS has gone and how confident you are in the signal. Performance rules read yesterday's settled data rather than real-time numbers, which means they evaluate against actual performance rather than intra-day noise. A creep-detection rule might look like: if ACoS exceeds target by 15% for two consecutive days, send an alert. That threshold and duration filter is what separates a meaningful signal from daily variance.
The rule does not replace your regular review. You still need to find and fix the source once the alert fires. What it does is watch for threshold breaches on the days between your check-ins, so the signal comes to you rather than waiting for your next scheduled review to surface it.
ACoS creep is a directional problem. It only becomes costly when the direction goes undetected long enough to compound. A disciplined weekly review plus an automated rule watching in between is what closes the gap. Each does something the other cannot.
Frequently asked questions
What is Amazon ACoS creep? ACoS creep is a sustained, gradual increase in advertising cost of sales across several weeks, with no single identifiable trigger. Unlike a spike, it does not show up on any one day. It accumulates until the cumulative rise becomes undeniable at month-end.
Why does my Amazon ACoS keep rising slowly? Gradual ACoS increases typically come from keyword drift, negative keyword gaps, rising competitor bids, slow conversion rate erosion, or budget running out earlier in the day. Most cases involve more than one factor operating simultaneously.
How often should I check Amazon ACoS to catch creep? Weekly, and always comparing the 7-day ACoS to your 28-day rolling average rather than just reading this week's number in isolation. The direction across 4 to 8 weeks is the signal. For coverage between reviews, a performance rule that alerts when ACoS crosses a set threshold is the automated backstop most accounts add once the manual review cadence is in place.
What is the difference between an ACoS spike and ACoS creep? A spike is sudden and usually traceable to a specific event. Creep is cumulative and has no single trigger. Spikes are visible in daily data. Creep only appears across a longer time horizon, which is why it goes undetected more often than spikes do.
Off Hours performance rules evaluate your ACoS against yesterday's settled data once daily, giving you a set of eyes between your manual reviews. Set a threshold, choose your action, and the rule watches for drift on the days you are not in the account. Start a free 14-day trial.