Most Amazon sellers run auto and broad match campaigns, glance at the search term report once in a while, and leave it there. That is the single biggest reason PPC accounts stall at a mediocre ACoS and never improve. The accounts that scale efficiently treat discovery campaigns as a keyword mine, not a destination — they systematically extract the converting search terms and promote them into dedicated, tightly controlled exact match campaigns. This process is called keyword harvesting, and it is one of the highest-leverage habits in Amazon PPC management.
This guide walks through exactly how StoreStride runs keyword harvesting for private label clients: what to pull, when to pull it, where it goes, and the mistakes that quietly waste ad spend.
Keyword harvesting in one sentence: let broad, phrase, and auto campaigns discover which exact search terms convert, then move those exact terms into their own exact match campaigns with independent budgets and bids — so you can control spend on your proven winners instead of letting them compete for scraps inside a discovery campaign.
Why Harvesting Matters
Auto, broad match, and phrase match campaigns exist to discover new, relevant search terms — they are research tools, not profit centres. Amazon's algorithm will match a broad keyword to dozens of related (and sometimes irrelevant) queries. Inside that noise are a small number of search terms that convert exceptionally well. If you leave them sitting inside the broad campaign, three problems compound:
- You cannot bid on them individually. A broad campaign has one bid for the whole keyword, so your best-converting search term and your worst-converting one get the same bid.
- Budget gets diluted. Your winning search terms compete for the same daily budget as the terms that are still being tested, so the proven performer can get budget-capped out by midday.
- You lose control of ACoS. You cannot push a high-converting term harder, or defend its ranking against competitors, if it is buried inside a discovery campaign.
Harvesting fixes all three. It moves your proven performers into a structure where you decide the bid, the budget, and the priority — and it keeps your discovery campaigns lean, so they keep doing their actual job of finding the next winner.
The Harvesting Flow
1. Discover
Auto, broad, and phrase campaigns serve ads against a wide net of customer search terms.
2. Identify
Pull the search term report and isolate terms that hit your conversion threshold.
3. Promote
Add the winning term as a new exact match keyword in a dedicated harvest campaign.
4. Exclude
Negative exact that same term in the source campaign so spend stops overlapping.
That loop repeats weekly for the life of the campaign. Over months, it transforms a broad, inefficient account into one built almost entirely from proven exact match keywords, with auto and broad campaigns running on small budgets purely to keep discovering the next candidate.
What Counts as a “Winner”: Setting Your Thresholds
You need firm, pre-agreed rules for what gets harvested — otherwise you will either promote too aggressively (cluttering your account with low-volume keywords) or too conservatively (leaving money on the table). Use data volume first, performance second:
| Signal | Minimum Threshold | Why |
|---|---|---|
| Clicks on the search term | 15–20+ clicks | Below this, conversion rate is statistical noise |
| Orders attributed to the term | 2+ orders | A single order can be a fluke; two or more shows a pattern |
| ACoS vs your target | At or below target ACoS | A term converting above target ACoS isn't a winner yet |
| Conversion rate vs account average | At or above average | Confirms the term is pulling its weight relative to the rest of the catalogue |
Terms that meet the click and order thresholds but sit slightly above target ACoS are still worth harvesting — moving them to exact match with a controlled bid often improves their ACoS, because you are no longer paying the inflated bid set for the whole broad match term.
How to Harvest: Step-by-Step
Pull the Search Term Report Weekly
Use Seller Central's Sponsored Products search term report, or Brand Analytics for a broader view. Filter to your discovery campaigns (auto, broad, phrase) and sort by orders, then by ACoS. A 7–14 day lookback is usually enough data for established listings; new listings may need 21–30 days to reach the click threshold.
Shortlist Candidates Against Your Thresholds
Apply the click, order, ACoS, and conversion rate thresholds above. Also sanity-check relevance — a converting search term that only makes sense for one specific use case of your product is still worth harvesting, but keep it in mind for ad copy and listing alignment.
Add as Exact Match in a Dedicated Harvest Campaign
Create (or use an existing) exact match campaign structured specifically to hold harvested terms — separate from your original manual exact campaigns if you want to track harvest performance independently. Start the bid at roughly the CPC that term was already paying inside the broad campaign, then adjust from real exact match data within a few days.
Negative Exact Match It in the Source Campaign
This is the step sellers forget, and it is the most important one. Add the exact search term as a negative exact in the auto, broad, or phrase campaign it came from. If you skip this, you are now bidding on the same term in two places, competing against yourself and inflating your own CPCs. See our negative keywords guide for the mechanics.
Monitor the Harvested Term Independently
Track its ACoS, conversion rate, and rank position on its own. A term that performed well inside a broad match can behave differently once it is the sole focus of a bid — usually better, since you can now defend its position with a precise bid rather than a blended one.
How Often Should You Harvest?
Weekly is the right cadence for most accounts. Daily is overkill and creates noisy, low-confidence decisions. Monthly is too slow — you leave proven winners starved of budget inside a discovery campaign for weeks while competitors can outbid you for the same term elsewhere.
- New or low-traffic listings: Bi-weekly or monthly, since it takes longer to accumulate the click volume needed for a confident decision.
- Established, high-traffic listings: Weekly, sometimes twice weekly during peak seasonal periods when search volume spikes and winners emerge faster.
- Post-launch (first 60 days): Weekly without exception — this is when the account structure is forming and harvesting compounds fastest.
Harvesting From Auto vs Broad/Phrase Campaigns
Both feed the same process, but they surface different things. See our guide on auto vs manual campaigns for the full structural comparison — for harvesting purposes, the key difference is this:
Auto Campaigns
Surface genuinely new search terms you would never have thought to target — adjacent use cases, substitute products, complementary items. Best source for discovering entirely new exact match keywords.
Broad & Phrase Campaigns
Surface variations, long-tail phrasing, and misspellings of keywords you already know are relevant. Best source for expanding an existing keyword into its highest-converting exact phrasing.
Common Harvesting Mistakes
- Forgetting the negative exact step. Without it, you bid against your own harvested keyword and inflate CPCs on both campaigns.
- Harvesting on too little data. One lucky order on a low-click term is not a pattern. Wait for the click threshold.
- Setting the exact match bid too low. Carrying over the broad match bid without adjustment often under-bids the term relative to the competition it now faces at the top of exact match search results.
- Never pruning the harvest campaign. Harvested terms can decay in performance over time (seasonality, new competitors, listing changes). Review harvested keywords quarterly, not just when adding new ones.
- Ignoring irrelevant but converting terms. If a search term converts but doesn't match your product well, it may be a sign your listing images or title are creating a mismatch with customer expectations — worth investigating before you scale spend on it.
Compounding effect: an account that harvests consistently for six months typically ends up with the majority of its spend sitting in exact match campaigns built entirely from proven, real customer search terms — not guesses from keyword research tools. That is why mature, well-run PPC accounts almost always have lower blended ACoS than newer ones with identical budgets.
Tools That Speed Up Harvesting
Harvesting can be done manually inside Seller Central with a spreadsheet, and many successful accounts run it exactly that way. For higher SKU counts, bulk operations and third-party software remove the manual grind:
- Bulk file uploads — see our bulk operations guide for adding harvested keywords and negatives across many campaigns at once.
- Rule-based PPC software (Helium 10, Perpetua, Teikametrics) — can automate the threshold check and even the promotion step, though human review of relevance is still worth doing before anything goes live.
- Search Query Performance report — complements the search term report with funnel-level data (impression share, click share) that can validate a harvesting decision beyond ACoS alone.
Is your account harvesting its best keywords — or losing them in broad match?
StoreStride builds and runs the full harvesting cycle for every client account: weekly search term reviews, promotion into exact match, and negative keyword hygiene to stop you bidding against yourself. Book a free audit and we'll show you exactly what your current structure is missing.
Get Free Amazon PPC Audit →Common Questions About Keyword Harvesting
Does harvesting work for Sponsored Brands and Sponsored Display too?
The principle is the same, but the mechanics differ. Sponsored Brands doesn't give you search term-level targeting the same way, so harvesting there usually means reviewing which headline/keyword combinations drive the most Sponsored Products-style conversions and feeding that insight back into your Sponsored Products structure. Sponsored Display harvesting is more about identifying which ASINs or audience segments convert, since it is not keyword-driven in the same way.
Should I pause the keyword in the broad campaign once harvested?
No — keep the broad or phrase keyword running. You are only negative-exacting the specific search term that converted, not the whole broad keyword. The broad keyword will continue surfacing other, different search terms you haven't harvested yet.
What if a harvested keyword's performance drops after promotion?
This can happen if the bid was set incorrectly, if a competitor has entered that exact match auction, or if the term was seasonal. Give it 7–10 days of exact match data before deciding whether to adjust the bid, pause it, or (rarely) remove the negative exact and let it fall back into the broad campaign.
How many keywords should a harvest campaign hold?
There's no fixed cap, but most well-run accounts group harvested keywords by theme (e.g. by use case or audience) into several harvest campaigns rather than one giant catch-all. This keeps budgets and bid strategies relevant to each group and makes performance easier to diagnose.