Auto campaign harvesting is the routine of using Sponsored Products automatic campaigns as a discovery feed, pulling converting shopper queries from the Search Term Report, and promoting those winners into manual campaigns while negatives prune the waste. Amazon Ads guidance frames this as the standard path from discovery to control, and tools can automate the repetitive parts of the cycle.
TL;DR:
- Harvesting should be done weekly or fortnightly, with higher frequency for high-spend campaigns to prevent wasted budget on unqualified terms.
- Promoted search terms should start in phrase match with moderate bids before tightening to exact match as data accrues.
- Only promote search terms or ASINs that meet minimum thresholds of 10 to 15 clicks and at least one or two conversions to minimize noise and overbidding.
- Automatically pulling and qualifying Search Term Reports with tools can save hours, but human judgment is needed for ASIN behaviour and listing changes.
- Irrelevant or unprofitable ASINs appearing in search reports should be negative targeted immediately to prevent wasted impressions and budget.
Table of Contents
- The harvesting funnel explained
- How to extract and qualify harvest candidates from the Search Term Report
- Promoting winners: how to move harvested terms into manual campaigns and set bids
- Monitoring, cadence and the metrics that prove harvesting is working
- Common pitfalls and practical fixes when harvesting from auto campaigns
- How automation and analytics speed safe harvesting
- The four-week harvesting routine worth starting with
- How Osellpa helps automate harvesting routines
- FAQ
- Sources
The harvesting funnel explained
Auto campaigns exist to find targets you haven’t thought of. Manual campaigns exist to bid on the targets you’ve proven work. Harvesting is simply the bridge between the two, and it follows a consistent shape every time you run it.
- Discovery: your automatic campaign serves ads against Amazon’s own matching logic across close match, loose match, substitutes and complements.
- Capture: you export the Search Term Report, which records every customer query and every ASIN your ad showed against.
- Qualify: you filter that report against clicks, conversions and spend to separate genuine winners from noise.
- Promote: qualifying search terms move into manual keyword campaigns, and qualifying ASINs move into manual product targeting.
- Prune: anything irrelevant or unprofitable gets added as a negative, in both the auto campaign and, where needed, the manual ones.
The reason this order matters is control. Amazon’s own targeting guidance recommends reviewing the automatic campaign’s Search Term Report specifically to identify strong keywords or products, then adding those targets to manual campaigns where bids can be set more competitively. Auto campaigns spend broadly and cheaply to learn; manual campaigns spend narrowly and deliberately to earn.
One distinction trips up a lot of sellers early on: not every row in the report is a search term. Some are ASINs, meaning your ad appeared on a competitor’s or a complementary product’s page rather than against a typed query. Shopper search terms belong in manual keyword campaigns. ASIN rows belong in manual product targeting campaigns, or in a negative list if they’re irrelevant. Treating the two the same way is one of the fastest routes to wasted spend.
How to extract and qualify harvest candidates from the Search Term Report
Before you touch bids, you need a clean shortlist. Export the report with these fields visible: customer search term, impressions, clicks, spend, conversions (orders), ACOS and, where available, ROAS. Sort by spend first, since that’s where the biggest wins and the biggest leaks tend to hide.
- Set a minimum click threshold (commonly 10 to 15 clicks) before judging a term at all.
- Require at least one or two conversions, or a conversion rate that beats your account average.
- Bucket by ACOS or ROAS against your breakeven point, not an arbitrary round number.
- Flag rows with high spend and zero conversions as negative candidates straight away.
- Separate ASIN-format rows from text search terms before applying any filter.
Cadence matters as much as the filter. Amazon’s marketing optimisation guidance points to weekly or fortnightly keyword reviews as a sensible rhythm, tightening for high-spend campaigns and loosening where conversion volume is naturally thin.
For ASIN entries specifically, the decision is binary: promote converting ASINs to a product targeting campaign, or add non-converting, irrelevant ASINs as negative targets so your auto campaign stops wasting impressions on them. Seller Central discussions confirm this ASIN-in-search-term behaviour is a known, recurring pattern in auto campaign reporting rather than an error.
Promoting winners: how to move harvested terms into manual campaigns and set bids
Once a term or ASIN clears your qualification bar, the promotion itself needs its own discipline. Jumping straight to exact match with an aggressive bid on a term that’s only proven itself over a handful of clicks is how harvesting backfires.
- Start promoted search terms in phrase match with a moderate bid, then tighten to exact once you have more data at the keyword level.
- Set a small trial budget for newly promoted keywords rather than folding them into an existing campaign’s shared budget.
- Promote converting ASINs into a dedicated product targeting campaign, bidding slightly below your average CPC until performance confirms itself.
- Add irrelevant or wasteful ASINs as negative product targets in the auto campaign the moment you spot them.
- Review promoted terms against the same qualification filter again after their first full cycle, not just once.
This staged approach mirrors what Amazon Ads itself recommends: promote to manual for tighter bid control, rather than letting the auto campaign keep absorbing spend on a term you already know converts.
Pro Tip: Cap your first bid on a newly promoted keyword at or below the CPC it achieved inside the auto campaign, then adjust once you have a week of standalone data.
Monitoring, cadence and the metrics that prove harvesting is working
A harvesting routine only earns its keep if you can see the result in numbers. Amazon’s optimisation guidance suggests weekly or fortnightly reviews as a baseline, shortening the interval for high-spend ASINs and lengthening it where traffic is light.
| Metric | What it tells you | Watch for |
|---|---|---|
| Conversions | Whether a term is a genuine winner or a fluke | Fewer than 2 to 3 orders is still noise |
| ACOS | Profitability of the spend behind a term | High ACOS despite conversions signals overbidding |
| ROAS | Revenue return relative to ad spend | Falling ROAS after promotion means the bid moved too fast |
| CTR | Whether the term actually matches shopper intent | Low CTR with decent conversions can still be worth keeping |
| Conversion rate | Reliability of the match beyond raw order count | Big swings week to week usually mean the sample is too small |
The most common false winner is a term with one or two conversions on very few clicks: it looks promising but hasn’t earned statistical confidence yet. The second is a term converting steadily but at an ACOS that quietly erodes margin, which only shows up once you look past the order count to the spend behind it.
Common pitfalls and practical fixes when harvesting from auto campaigns
Most harvesting failures come from skipping a step rather than from the process itself being flawed.
- Promoting a raw report row verbatim, including filler words or plural variants, instead of normalising it into a clean keyword with the right match type.
- Treating an ASIN row as if it were a typed search term, which routes it into the wrong campaign type entirely.
- Promoting a term after a handful of clicks rather than waiting for a sample size large enough to trust.
- Forgetting to negative the auto campaign once a term has been promoted, which causes both campaigns to compete for the same query.
- Letting a promoted keyword sit untouched after its first bid, rather than feeding its new performance back into the next review.
How automation and analytics speed safe harvesting
Running this routine by hand every week is possible, but it’s also where sellers quietly stop doing it. Some tools integrate directly with Amazon’s API to automate the report pulls and apply qualification rules consistently rather than relying on memory or a spreadsheet template that drifts over time.
Scheduled reviews, automatic flagging of promotion candidates and safe negative application all run in the background, cutting the manual hours harvesting normally demands. What automation shouldn’t replace is judgement on anything unusual: a sudden ASIN spike, a listing change mid-cycle, or a creative decision about which keywords belong in your next copy refresh still need a human look.
The four-week harvesting routine worth starting with
In weeks one and two, let the auto campaign run uninterrupted so Amazon’s matching has time to learn, while you set up baseline reporting alongside it. Week three is your first real qualification pass: pull the report, apply your filters, and promote the strongest candidates into small-budget manual tests. By week four, scale what’s proven, negative what isn’t, and lock in a weekly or fortnightly review from there on.
— Harry
How Osellpa helps automate harvesting routines
Running this funnel manually works, but it eats hours every week that most sellers would rather spend on sourcing or listings. We built our PPC advertising tools to handle the repetitive parts: automated Search Term Report pulls, qualification rules applied the same way every cycle, and bid suggestions for newly promoted keywords.
- Automatic API-based report pulls, so nothing waits on a manual export.
- Consistent qualification rules applied to every review, not just the ones you remember to run.
- Bid guidance for promoted keywords and ASINs based on their proven performance.
We still recommend a human glance at anything unusual the system flags. You can see current plans, including Launch at £10 a month, on our pricing page.
FAQ
What is auto campaign harvesting on Amazon?
Auto campaign harvesting is the practice of using Sponsored Products automatic campaigns to discover converting search terms and ASINs, then promoting the strongest performers into manual campaigns for tighter bid control. Amazon Ads guidance describes this discovery-to-manual path as the intended use of automatic targeting.
How often should I review and harvest Search Term Reports?
A weekly or fortnightly cadence works for most accounts, with shorter intervals for high-spend campaigns and longer ones where conversion volume is naturally low, as Amazon’s optimisation guidance suggests. Reviewing too infrequently lets budget-draining terms run unchecked for longer than necessary.
Why do ASINs appear in my Search Term Report instead of keywords?
ASIN entries mean your ad appeared on a product detail page rather than against a typed search query, a pattern confirmed in Seller Central discussions on the topic. Treat converting ASINs as product targeting candidates and irrelevant ones as negatives, never as keywords.
What minimum data should I see before promoting a keyword?
Most sellers wait for at least 10 to 15 clicks and one or two conversions before trusting a term enough to promote it, since smaller samples are prone to noise. Starting the promoted term in phrase match with a modest trial budget protects against overcommitting on an early result.
Can automation handle the whole harvesting process for me?
Automation can reliably handle the repetitive parts, including report pulls, rule-based qualification and negative application, which is what tools like Kingpin’s AI-powered sales agency are built for. Human review still matters for unusual ASIN behaviour, listing changes and creative decisions that numbers alone can’t resolve.
Sources
For primary guidance, start with Amazon’s targeting guide and the Ads API reporting docs. Seller Central’s ASIN-in-search-term thread is worth reading for real examples, and our own search term discovery guide expands on the qualification step in more depth.
