You harvest keywords — but your auto and category campaigns also show which competitor ASINs convert for you. The promote-negate-isolate funnel for product targeting, with real numbers from my own seller account.
I harvest keywords religiously. Every week I pull the search-term report, promote the converters into exact-match campaigns, and clean up the waste. It is the highest-leverage habit in my PPC routine.
Then I looked at my own account from the other side. 3,968 product targets — and not one was a deliberate, exact-ASIN harvest. They were auto-targeting clauses and Amazon's "expanded" guesses: placements the system chose, not ones I had proven and promoted. The data on which competitor products convert for me was sitting right there in my reports. I just was not acting on it. (Those are real numbers from my own Rowdy Rooster Woodworks account, pulled the day I wrote this — not a hypothetical.)
If you harvest keywords but ignore ASINs, you are doing the same thing. Here is the other half of the harvest.
What an ASIN harvest actually is
When your ad shows on a product's detail page and someone buys, that ASIN is a proven placement — the same way a converting search term is a proven query. Your auto campaigns (through the "complements" and "substitutes" clauses) and your category campaigns constantly place your ad on other products' pages. Some convert. Some convert better than your keywords.
Promoting one means creating a product target — Amazon calls it ASIN_SAME_AS — where you bid that exact page deliberately. It is keyword harvesting's twin: search queries become exact keywords; converting product pages become exact ASIN targets. Most sellers skip it because the ASIN shows up as a cryptic B0… string instead of readable words, so it gets ignored. That is the whole edge — proven detail-page real estate your competitors are ignoring in the same report.
The funnel: promote, negate, isolate (for products)
If you have harvested keywords, this is muscle memory — just pointed at ASINs.
-
Promote. Create the converting ASIN as an exact product target (
ASIN_SAME_AS) in a dedicated product-targeting performance campaign. Set the bid at the placement's observed cost-per-click, not a guess. One ASIN, controlled bid, isolated so you can read its numbers.
-
Negate it in the source. Add the ASIN as a negative product target in the source auto or category campaign so the two stop bidding against each other. The detail that trips people up: Amazon has no campaign-level negative ASIN — negative product targets exist only at the ad-group level, so the negation goes on the specific ad group the placement came from. (We hit this building the feature; the API will not accept it any other way.)
-
Isolate. Now exactly one campaign bids that product page, at a bid you set on purpose. The discovery campaign keeps finding the next one; the performance campaign scales the proven one.
Exact, never "expanded"
Amazon gives you two flavors of ASIN target: ASIN_SAME_AS (this exact product) and ASIN_EXPANDED_FROM (this product plus others it thinks are similar). For harvesting, always use exact. The point is to scale a proven winner — you have evidence this page converts. Expanded throws your bid at a cloud of look-alikes Amazon picks, riding on the back of your one real result. That is re-introducing the guesswork you just escaped. Expanded is a discovery tool; harvesting is what comes after discovery. (My account had 225 expanded targets and zero exact ones — exactly backwards.)
When is an ASIN actually ready?
Same discipline as keywords. The bar is not "it got a sale once."
-
Orders ≥ 2 (the floor, not the goal). Two independent conversions is the minimum for "real pattern, not fluke." Treat the promotion as a test the isolated campaign will confirm or reject.
-
ACoS ≤ your break-even. Break-even ACoS = (margin after fees ÷ price) × 100. Promoting an unprofitable placement to exact does not fix the math — it just gives you a cleaner bid on a loser. If a lower exact bid cannot get it under break-even, it is not a winner.
-
The rescue band. If a placement converts but sits a little over your target ACoS, exact is exactly how you fix it — you take control of the bid and pull the ACoS down, instead of letting an auto clause overpay.
-
Lookback ≥ 14 days. Detail-page attribution lags like everything else on Amazon. Two weeks is the minimum before the numbers are trustworthy.
Automate the busywork, keep the control
The discovery and the bookkeeping can be automated; the decision should not be. The work that kills this habit is manual: scanning reports for B0… strings, checking each one's orders and ACoS, creating the target, remembering to add the matching negative in the right ad group, recording what you did so you do not re-harvest it next week.
That is exactly what we automated in RedHen Labs. It reads your reports, finds the converting ASINs that clear your thresholds, and queues two changes together — the exact product target in your performance campaign and the negative ASIN in the source ad group — then stops and waits for your one-click approval. Nothing ships to Amazon until you say so. The software does the tedious part; you keep the judgment. Try it free for 14 days.
The takeaway
The waste and the opportunity on Amazon are usually the same thing: invisible, not complicated. I had harvested keywords for years and still had 3,968 product targets without a single deliberate ASIN harvest in the pile. Once I could see it, the fix took minutes. If you only harvest keywords, you are leaving the converting-product half of your own data on the table. Pull your product-targeting and auto search-term reports, find the B0… placements that converted, and promote the proven ones to exact. That is the whole move.