Most account managers open the search term report, scan for a few obviously bad terms, add them as negatives, and close the tab. That is not analysis. That is housekeeping. The report is the single richest source of customer intent data you get from Amazon, and treating it as a chore instead of a strategic input is why so many accounts plateau at the same ACoS for years.

The search term report tells you, in plain language, what shoppers typed before they clicked your ad and bought (or didn't). Every row is a small experiment Amazon already ran for you. Your job is to read the results and make four decisions: harvest it, negate it, promote it, or watch it. Here is how to do that with discipline instead of guesswork.

Start With the Right Time Window and the Right Filter

Before you make a single decision, get the window right. Pull a minimum of 30 days of data for most catalogs, and go to 60 if your daily click volume is thin. A search term with three clicks and one sale over a week tells you nothing. The same term with forty clicks and six sales over a month tells you something real.

Filter out noise before you filter for insight. Sort by spend first, not by ACoS. A term with a terrifying 180% ACoS on $4 of spend is irrelevant. A term with a mediocre 45% ACoS on $600 of spend is where your week should go. This is the same discipline we cover in setting a target ACoS for each product: the number only means something next to the dollars behind it.

Harvest: Move Winners Into Their Own Campaign

A search term earns harvest status when it has converted at or below your target ACoS with enough volume to trust the number, typically ten or more clicks with at least one sale, repeated across the data window. These terms do not belong buried inside a broad or phrase match campaign anymore. Pull them into an exact match campaign you control directly.

Why bother, if they're already converting? Because inside an automatic or broad campaign, Amazon's algorithm is still deciding how much of your budget that term deserves relative to every other term it's testing. In its own exact match campaign, you set the bid, you control the budget, and you can push it harder without dragging a dozen unrelated search terms along for the ride. This is also where you protect your best performers from bid changes made for the sake of the campaign average.

A converting search term buried in a broad campaign is a signal you're ignoring, not a result you're capturing.

Negate: Cut the Terms That Will Never Convert

Negation is the fastest profit lever in the entire report and most sellers under-use it. A term earns a negative when it has real spend (again, use your volume threshold) and zero sales, especially when the term reveals clear mismatched intent, like a shopper searching for a different size, a competitor brand name, or a use case your product doesn't serve.

Add these as negative exact matches at the ad group or campaign level, depending on where the waste is showing up. If the irrelevant term is a broader pattern (a completely different product category, say), consider a negative phrase to close the whole family of searches rather than whacking one term at a time. We go deeper on the mechanics of this in the cheapest profit on Amazon, but the short version is: every dollar spent on a term that has never converted after real volume is a dollar you've already decided not to collect.

One caution: don't negate on ACoS alone. A term with a high ACoS but at least one sale is a candidate for a bid cut, not necessarily a negative. Negation is for terms that prove, through volume, that they will not convert, not terms that converted expensively once.

Promote: Bid Up on Terms Showing Momentum

Promotion is the decision most reports skip entirely because it requires judgment rather than a clean threshold. Look for search terms with a strong recent trend: rising click-through, a conversion rate climbing toward or past your average, and ACoS trending down even if it isn't there yet. These are terms worth a bid increase before they've fully proven out, because waiting for perfect data means you lose the ranking momentum a strong term can build.

This matters even more for newer products still building rank, where capturing an early-trending term cheaply compounds over weeks. If you're running this analysis on a product still inside its first month, cross-reference it against the keyword strategy in the first 30 days of an Amazon launch, since a search term converting well in week two often deserves a bigger role in week four than your original launch plan assumed.

Don't Confuse Promotion With Chasing Volume

Promoting a term because it has impressions is not the same as promoting a term because it converts. Keep the bar for promotion tied to conversion rate and ACoS trend, not raw traffic. High-impression, low-conversion terms are watch candidates, not promotion candidates, no matter how good they look in a volume-sorted view.

Watch: The Category Everyone Forgets

Watch terms are the ones with real spend but not enough data yet to confidently harvest, negate, or promote. This happens constantly with new broad and auto campaigns, where a term might have four clicks and one sale this week. That's not a result, that's a sample size problem.

Log these in a running tracker and revisit them on your next report pull. The mistake most accounts make is forcing a decision on a watch-tier term out of impatience, either negating it too early and losing a potential winner, or promoting it on a single lucky sale. Give it another data cycle. If it's still ambiguous after two or three pulls, that ambiguity is itself information: the term probably isn't strong enough to matter either way, and you can let it ride at a neutral bid.

Where This Fits Into the Bigger Picture

None of this works as a one-time cleanup. The search term report should be read on the same cadence every week or two, not whenever ACoS spikes and someone panics. A single pass tells you what happened last month. A repeatable system tells you what's changing right now, which is the only version of this data that actually improves your account.

If you're also running Sponsored Brands or Sponsored Display alongside Sponsored Products, note that search term data behaves differently across those formats, something worth factoring in if you're deciding where each ad dollar works hardest.

What to Do This Week

Pull your last 30 days of search term data today. Sort by spend, not ACoS. Run every row through the four buckets: harvest anything converting at or below target ACoS with real volume, negate anything with spend and zero sales, promote anything trending up in conversion, and log everything else to watch. Do it again in two weeks. The report doesn't change your account. The decisions you make from it do.