Halfway through your second cup of coffee, you spot something interesting. An auto campaign you set up six months ago for an air fryer cookbook has been quietly producing orders on a phrase you didn't see coming: "easy air fryer recipes." Eleven orders over 30 days. ACoS 14%. Your target is 35%.
You'd never targeted that phrase deliberately. The auto campaign's broad matching surfaced it. And now it's outperforming everything else in the campaign by a factor of two.
You run the bulk sheet through BulkSheet Pro. The Harvest Terms page lists it among 28 other terms with the same story: proven converters discovered by auto and broad campaigns, now ready for their own dedicated structure.
This is what BulkSheet Pro does next.
What the Harvest Terms page shows you
The Harvest Terms page surfaces a broad universe of candidates. By default it shows every search term in your account that has produced some commercial activity — at least one sale or one order — and that isn't already in an exact or negative-exact match ad group. The page sorts them best-first using a scoring function that prioritizes terms with low ACoS and real order volume, so the highest-leverage candidates float to the top.
There's also a Best CVR view that re-sorts the same universe by conversion rate descending — useful when you want to spot terms converting exceptionally well even if absolute order volume is still small.
From any view, you can manually select terms to harvest. The page is the broad surface; you decide what goes into the workflow.
What BulkSheet Pro recommends automatically
The Rec Actions filter narrows the page to a smaller, stricter set: terms the rec engine has identified as worth committing to a dedicated campaign on their own merit. The rec engine isn't looking for any term that converted — it's looking for terms that have crossed a threshold of proof.
A term qualifies for the automated recommendation if all of these are true:
It's in an ad group whose match type isn't already exact (so Auto, Broad, Phrase, Modified Broad, Product Target ad groups are all in play — anywhere a search term can surface that wasn't deliberately targeted as exact).
It has at least two orders. The two-order floor matters because a single order can be a fluke. Two orders within the same window is a pattern starting.
Its ACoS is at or below half the target ACoS for its SKU and campaign type. Not just at target — significantly outperforming it. The reasoning: harvesting commits to giving the term its own structure with its own bid and budget. That commitment is worth making only when the term has proven itself well past the breakeven point.
The half-target rule is the gate that separates "harvest now" from "let it keep performing where it is." A term at 90% of target is fine where it is — you're getting the orders, the auto campaign is doing its job. A term at 30% of target is leaving money on the table; it's converting so well that the parent campaign's bidding is probably losing impressions on the floor.
What confidence actually means here
Once a term qualifies, BulkSheet Pro assigns confidence based on orders alone:
- 5+ orders: HIGH confidence. Statistically a real pattern.
- 3–4 orders: MEDIUM confidence. Likely a pattern, worth promoting.
- 2 orders: LOW confidence. Met the minimum, but treat it as exploratory.
By default, only HIGH confidence terms pre-select into the harvest queue when you switch the page to Rec Actions. The medium and low ones are visible — you can opt into them — but BulkSheet Pro is intentionally cautious about committing them to permanent structure on thin evidence.
What harvesting actually does
When you harvest a term, BulkSheet Pro builds a new campaign around it. Not a tweak to an existing ad group — a fully scaffolded new campaign. The structure includes:
- A campaign name that follows a SKU + match type + term convention, so your account stays organized as it grows
- A single ad group dedicated to the harvested term
- The keyword itself, set to the match type or types you chose — Exact, Phrase, Broad, or Modified Broad. You can pick more than one in the same workflow; BulkSheet Pro will scaffold a separate campaign for each match type
- Your bid, your daily budget, your start date, your portfolio assignment
- Placement multipliers for Top of Search, Product Pages, and Rest of Search — set on the harvest form, so you can bias the new campaign toward where you want it to compete
All of this gets queued in pending changes, ready for one-click export to Amazon's bulk operations sheet. The 28 terms in this morning's session can each be customized at the row level — different SKU, different bid, different budget, different start date — or you can use the form-level defaults and ship them all at once.
The companion negative
One thing the Harvest workflow doesn't do automatically is negate the term from its origin. This is intentional — but it's also the move you'll want to make.
If you harvest "easy air fryer recipes" into a new exact-match campaign while leaving the originating auto campaign untouched, both campaigns will compete for the same impression. You'll be bidding against yourself in Amazon's auction, and the impression will route to whichever campaign has the higher placement-adjusted bid for that surface. Either you lose control of which campaign earns the conversion, or you start cannibalizing your own budgets.
BulkSheet Pro handles this through a paired workflow. The Search Terms page (with cross-campaign negation) and the Harvest Terms page are designed to be used together. Most experienced operators run them in sequence: harvest the converters, then negate the harvested terms from their original ad groups in the same export. Both moves go out as part of the same bulk sheet upload, and the new exact-match campaign takes over cleanly the next time Amazon ingests it.
This is the part many tools miss. Identifying winners is easy. Promoting them without leaving a mess behind is the harder half of the workflow.
What BulkSheet Pro won't do
It won't harvest weak signals. The 2-order minimum and the 50%-of-target ACoS gate exist because committing to a new campaign for a thin signal creates account complexity without revenue. BulkSheet Pro defaults to LOW-confidence treatment of borderline cases and keeps them out of the default Rec Actions filter.
It won't harvest exact-match terms. They're already in a dedicated structure where you control the bid and budget. Harvest is for terms that haven't been promoted yet — terms still surfacing through auto, broad, or phrase ad groups where Amazon's matcher decides what to bid on.
It won't pretend match type doesn't matter. Harvest can promote a term into Exact, Phrase, Broad, or Modified Broad — and you can pick more than one in the same workflow, which creates a separate campaign per match type. Exact is the precision instrument; the wider match types shift the campaign's role from "compete on this specific phrase" to "compete on this neighborhood." There's no default — you pick at least one before the form will commit, and the choice is yours.
The export
You hit Export. The bulk sheet contains 12 new campaign rows (the HIGH-confidence harvest queue, post-review), 12 corresponding ad group rows, 12 keyword rows, plus the 12 cross-campaign negatives you queued separately on the Search Terms page to keep your new campaigns from competing with their origins.
You upload to Amazon, wait for ingestion, and the new campaigns start running by tomorrow morning. The originating auto campaign is routed away from "easy air fryer recipes" by the new negative. The dedicated campaign picks up the impressions with a bid calibrated to your target ACoS.
A week later you check the harvested campaign's first reports. Eleven orders became fourteen, sixteen, eighteen. ACoS held at around 16%. The new campaign is doing what the auto campaign was doing — but with you in control of the bid and the budget instead of Amazon's broad matcher.
The point isn't that BulkSheet Pro found 12 great keywords. It's that it found the 12 keywords that had earned the right to be promoted, and gave each one the structural commitment it needed to keep working.