Amazon PPC keyword research

Sources of keyword ideas

Relevance before volume

A high-volume search is worthless if shoppers searching it want something different. 'Water bottle' and 'glass water bottle' attract different buyers; a steel bottle advertised on the second wastes clicks. Ask of each keyword: would a shopper typing this be pleased to see our product?

Match types

Amazon's manual keyword targeting offers broad, phrase and exact match. Broad and phrase reach variations and help discover new searches; exact match targets a term and close variations, giving the most control. Researched terms with uncertain value often start in phrase or broad; proven terms move to exact.

Harvesting from real searches

Search term reports list the searches that triggered your ads and their results. Terms that convert at an acceptable cost become new keywords; terms that spend without converting become negatives. Repeating this regularly is the most reliable form of keyword research once campaigns are running.

A keyword research routine

1. 01

   Seed

   Core terms from product and listing.

   Output: Seed list

2. 02

   Expand

   Suggestions, competitors, auto campaigns.

   Output: Candidate list

3. 03

   Filter

   Relevance and intent checks.

   Output: Approved keywords

4. 04

   Group

   By intent, product and match type.

   Output: Ad group plan

5. 05

   Harvest

   From search term reports, regularly.

   Output: Updated keywords and negatives

Example: researching keywords for a yoga mat

Keyword research checks

Grouping by intent

Group keywords by what the shopper wants: generic category terms, specific feature terms, problem-based terms and brand terms. Each group tends to convert differently and deserves its own bids and reporting.

Keywords and the listing

Research also shows how shoppers describe your product. If important terms are missing from your title or bullets, adding them where accurate can help shoppers recognise the product and may support relevance.

Specific searches

Longer, more specific searches usually have less volume but can convert well, because the shopper knows what they want. They are often worth testing alongside broad category terms.

Third-party tools

Keyword tools estimate search volumes and competition. Treat their figures as estimates and confirm value with your own campaign data.

Writing down what the product is not

Before expanding a seed list, note the features your product lacks: a size it does not come in, a material it is not made from, an age group it is not suitable for. That short list makes filtering faster and becomes the starting point for negative keywords. It also stops attractive-looking searches slipping through simply because they share a word with your product.

Brand and competitor terms

Searches for your own brand name often come from shoppers already looking for you, so they behave differently from category searches and are best kept in their own group. Bidding on competitor brand names is a separate decision: it can reach shoppers comparing options, but relevance may be weaker. Check Amazon's current advertising policies before targeting other brands' names, and report these terms separately so their results do not blur your category figures.

How often to revisit research

Shopper language changes with seasons, trends and new products in the category. A review of search term reports on a regular schedule, with a fuller research refresh when you launch variants or enter a new marketplace, keeps keywords current without constant rework.