Amazon Search Term Harvesting Explained

Search term harvesting is the process of finding search terms that convert in one campaign and promoting them to another campaign where you can control the bid more precisely. It is one of the most common Amazon PPC optimisation workflows, and it exists because Amazon's auto campaigns are good at discovering what shoppers search for but give you limited control over how much you bid on each term.

How harvesting works

The basic workflow has three steps:

1. Discovery: An auto campaign (or a broad/phrase match manual campaign) runs and matches your product against a wide range of shopper search terms. Amazon decides which searches trigger your ad.
2. Identification: You review the Search Term Report to find terms that resulted in sales. These are your harvesting candidates: search terms that real shoppers typed, that matched your product, and that led to a purchase.
3. Promotion: You add the converting search term as an exact match keyword (or phrase match, depending on your strategy) in a manual campaign. You set a specific bid for that term. Then you add the term as a negative exact in the original campaign so traffic for that term flows through your manual campaign instead.

The result is that your best-performing search terms get their own bids, separate from the auto campaign's algorithmic bidding. You can increase bids on high-value terms and decrease them on marginal ones, which you cannot do inside an auto campaign.

Why the Search Term Report is the data source

The Search Term Report (available in Seller Central under Advertising, then Reports) shows you the actual search terms that triggered your ads, along with impressions, clicks, spend, sales, and orders for each term. This is the raw data behind any harvesting decision.

Without this report, you are guessing. The campaign dashboard shows you keyword-level performance, but it does not show which specific search terms matched against those keywords. A broad match keyword like "water bottle" might match dozens of search terms, some profitable and some not. The Search Term Report tells you which ones.

The three common harvesting modes

Whether you do this manually or use a tool, the workflow generally falls into one of three modes:

1. Fully automatic

A tool monitors the Search Term Report continuously. When a search term meets predefined criteria (for example, two or more orders and an ACoS below your target), the tool automatically adds it as a keyword to your manual campaign and negates it in the source campaign. No human reviews the decision.

The advantage is speed and consistency. Every qualifying term gets harvested without waiting for someone to run a report. The risk is that the criteria are blunt: they cannot account for context like seasonality, brand strategy, or whether a term is about to become irrelevant.

2. Semi-automatic with thresholds

The tool identifies candidates that meet your thresholds and presents them to you for approval. You review the list, approve or reject each term, and the tool executes the approved changes. This is the middle ground: you get the efficiency of automated identification with the safety of human judgement before anything changes.

Most sellers find this is the most practical mode. The tool does the tedious work of scanning reports and filtering by criteria, and you spend your time on the decisions that benefit from context.

3. Manual review queue

You download the Search Term Report, filter and sort it yourself, identify candidates, and make changes in Campaign Manager or via bulk upload. There is no automation. This is the slowest approach, but it gives you complete control and complete understanding of what is happening in your campaigns.

Manual harvesting makes sense when you are learning how your campaigns work, when your ad spend is low enough that the volume of search terms is manageable, or when you want to understand the data before trusting any automation with it.

When manual approval matters

Fully automatic harvesting is tempting because it removes a task from your weekly workflow. But there are situations where a human should review the decision:

Branded terms

If a competitor's brand name appears as a converting search term, you need to decide whether to harvest it deliberately or leave it alone. Bidding on competitor brand terms is allowed on Amazon, but it is a strategic decision with implications for your ACoS and competitor response. An automated rule does not weigh those considerations.

Low-volume converters

A search term with one order might look like a conversion, but one order is a small sample. It might be a coincidence. Automated systems with a threshold of "one or more orders" will harvest aggressively, creating a manual campaign full of keywords that each converted once and may never convert again. A human can assess whether the term makes logical sense for the product, not just whether a number crossed a line.

Seasonal terms

A search term that converts in November ("christmas gift for dad") should not necessarily be a permanent keyword in your manual campaign. Harvesting it makes sense during the season, but it needs to be paused or removed afterwards. Automated systems do not inherently know which terms are seasonal unless you build that logic in.

Terms that need a different campaign structure

Sometimes a converting search term reveals a new audience or use case for your product. Rather than adding it to an existing manual campaign, it might warrant its own campaign with tailored ad copy and a separate budget. A human recognises this; an automated rule just adds the keyword to the designated destination campaign.

Setting up a harvesting workflow

Whether you automate or not, the fundamentals are the same:

1. Run auto and broad campaigns long enough to gather data. You need search terms with meaningful click and conversion counts. A campaign that has been running for two days does not have enough data. Four to six weeks is a reasonable starting point for most products.
2. Define your harvesting criteria. What makes a search term worth promoting? Common criteria include: minimum number of orders (two or three), ACoS below your target, and a minimum number of clicks (to ensure the conversion rate is based on a reasonable sample).
3. Set up your destination campaign. Create a manual exact match campaign (or an existing one) where harvested keywords will go. Give it a budget and default bid that make sense for your margins.
4. Review and promote. Apply your criteria to the Search Term Report, add qualifying terms to the manual campaign, and negate them in the source campaign.
5. Monitor the harvested keywords. A term that converted in an auto campaign may perform differently in a manual campaign at a different bid. Check performance after two to four weeks and adjust bids or pause underperformers.

Common harvesting pitfalls

• Forgetting to negate: If you promote a term to a manual campaign but do not negate it in the source campaign, you end up bidding on the same term in two places. This can increase your costs without improving performance.
• Harvesting too aggressively: Moving every search term with a single conversion into a manual campaign creates a bloated keyword list full of terms that may not perform. Be selective.
• Never reviewing harvested keywords: A keyword that was profitable when harvested may stop converting over time. Harvested keywords need the same ongoing review as any other keyword in your campaigns.
• Ignoring match type in the destination: Most sellers harvest into exact match for maximum control. If you harvest into broad or phrase match in the manual campaign, you may end up matching the same irrelevant terms you were trying to escape.

Harvesting and listing quality

Harvesting works best when your listings are already strong. A search term that converts well does so partly because of your ad, but also because of your listing: the title, images, price, reviews, and bullet points that convince the shopper after they click. If your listing is weak, harvesting converting terms into a manual campaign will not fix the underlying conversion problem. Stronger listings are covered in our backend search terms guide and the title limit guide.