Most brand owners open the search term report, feel their eyes glaze over at row 400, and close it. That is the moment they hand a real advantage back to competitors who don't. The search term report is the single most honest document in your Amazon account. It tells you, in plain text, exactly what shoppers typed before they gave you money or wasted your ad spend. The problem isn't the data. It's that nobody taught you how to sort it into decisions.
This is a walkthrough, not a theory piece. By the end, you will have a repeatable weekly process that turns rows of search terms into four buckets: harvest, negate, promote, and watch.
Start With the Right Time Window and the Right Columns
Pull the report at the search term level, not the placement or targeting level, and set the window to at least 30 days for an established product, 60 to 90 if you have lower volume. Shorter windows drown you in noise from single-click anomalies.
You need six columns on screen at once: search term, campaign/ad group, impressions, clicks, spend, orders, and ACoS (or spend divided by sales, if you prefer to calculate it yourself). Everything below depends on having all six visible together, because a search term with high spend and zero orders reads completely differently than one with high spend and strong orders. Isolating any single column is how brand owners talk themselves into bad decisions.
Sort by spend, descending. That single sort does more work than any filter you could apply. It puts your biggest risks and your biggest wins at the top of the list, in the same place a strategist would look first.
Bucket One: Harvest
A search term earns a harvest decision when it has generated at least one order (ideally two or three, if volume allows) at an ACoS at or below your target ACoS for that product, and it is not already an exact-match keyword somewhere in your account.
Harvesting means adding that exact term as its own exact-match keyword in a dedicated ad group, with a bid you control directly, rather than leaving it to ride inside a broad or phrase match campaign where Amazon's algorithm decides how much of your budget it gets. This is the single highest-leverage move in the whole exercise. It takes proven demand and gives you a steering wheel.
Check your product listing before you harvest anything. If a term is converting from ads but isn't reflected anywhere in your title or bullets, that is a signal worth acting on beyond PPC. It may point to a gap covered in how to write Amazon titles that rank and still read like English, or it may belong in backend fields as discussed in our piece on backend keywords and indexing.
Bucket Two: Negate
A search term earns a negation when it has spent meaningfully (set a floor, for example twice your target CPC, or a fixed dollar amount like $15, depending on your account size) with zero orders, or when it is simply irrelevant to what you sell. Irrelevant terms show up more often than you'd expect, especially in broad match campaigns where Amazon's matching logic stretches further than most sellers realize.
Negate at the level that matches the mismatch. If the term will never convert for this product under any campaign, negative exact it across the account. If it's only wrong for this particular ad group but might work in another, negative exact it at the ad group level only. Getting this distinction wrong is how sellers accidentally choke off a term that would have converted somewhere else in the account.
Every dollar spent on a search term that has already proven it won't convert is a dollar stolen from one that will.
Do this weekly, not monthly. Waste compounds fast in a growing account, and a term that burns $20 a week burns over $1,000 a year if it sits there unaddressed. This is the same discipline covered in more depth in negative keywords: the cheapest profit on Amazon, and it pairs directly with the ACoS management approach in scaling PPC without letting ACoS run away.
Bucket Three: Promote
Promote is different from harvest. Harvest is about giving a proven term its own home. Promote is about giving more budget or a higher bid to a term or ad group that is already outperforming, before it hits a budget ceiling or gets starved by the auction.
Look for terms with a low ACoS relative to target, decent order volume, and impression share that suggests you're not winning every available auction. If a term is converting at half your target ACoS but only shows up on a handful of impressions relative to its search volume, you are leaving profitable sales on the table. Raise the bid in controlled increments, no more than 10 to 15 percent at a time, and recheck in a week rather than making one large jump and losing your ability to attribute the change.
This is also where Sponsored Brands earns a look for high-intent branded or near-branded terms that are converting well in Sponsored Products. The tradeoffs between the two formats are covered in Sponsored Brands vs Sponsored Products: where each dollar works hardest.
Bucket Four: Watch
Not every term deserves an immediate decision, and treating every row as urgent is how strategists burn out and stop doing this altogether. Watch is for terms with too little data to act on confidently: one or two clicks, no orders yet, spend under your negation floor. Tag these and revisit them next week rather than pretending you have a verdict today.
Watch is also where dayparting questions live. If a term looks weak in aggregate but you suspect its performance shifts by time of day, that's worth investigating before you touch the bid, a question we walk through in dayparting on Amazon: should you really bid by hour of day.
Building the Weekly Habit
The report itself takes 20 to 30 minutes to work through once you know the four buckets. The discipline is doing it every week, on the same day, before spend has a chance to compound in the wrong direction. Skipping a week doesn't just delay a decision, it lets a bad term keep spending and a good term keep getting starved.
Keep a running log of what you harvested, negated, promoted, and moved to watch. Over a few months, that log becomes a map of exactly which terms drive your business and which ones you've already ruled out, so you stop re-litigating the same decisions every quarter.
What to Do This Week
Pull your last 30 days of search term data today. Sort by spend. Work the top 50 rows into the four buckets before you do anything else in the account this week. That single pass will surface more immediate ACoS improvement than most sellers get from a full campaign restructure, because it targets the exact terms where your money is already landing.