Amazon PPC keyword research services

How a keyword research engagement runs

1. 01

   Brief

   Product, audience, what it is not.

   Output: Product notes

2. 02

   Gather

   Sources including your search terms.

   Output: Candidate list

3. 03

   Filter

   Relevance and intent.

   Output: Approved list

4. 04

   Organise

   Groups, match types, priorities.

   Output: Campaign-ready sheet

5. 05

   Handover

   Negatives and listing suggestions.

   Output: Final deliverable

A good brief

Tell the service what the product is, who buys it and what it is not. 'Not suitable for children' or 'not foldable' helps rule out irrelevant searches. Share search term reports from existing campaigns, which are more valuable than any tool estimate.

What a useful deliverable contains

Warning signs

• Huge unfiltered lists

  Irrelevant terms waste spend.

• Volume without relevance

  Clicks from the wrong shoppers.

• Ignores your search term data

  Misses the best evidence.

• No negatives suggested

  Only half the job.

Relevance checks

Ask how the service checks relevance: by reviewing search results for each term, comparing products shown with yours, or using your conversion data. The method should be explained, not hidden behind a tool.

Example: a keyword research deliverable

What you should receive

Volume estimates

If the deliverable includes search volumes from tools, treat them as estimates. Your own campaign data will show real performance.

One-off or ongoing

Initial research helps you launch; ongoing harvesting from search term reports keeps keywords current. Ask whether the service offers both, and whether ongoing work is part of PPC management.

What to agree

Agree products in scope, sources used, deliverable format, number of revisions and whether campaign setup is included.

Your part in the work

A service can only filter well if it understands the product. Expect to answer questions about features, materials, sizes and the customers you are trying to reach, and to review a draft list before it is finalised. You know the product better than any outside researcher, so a short review of borderline terms can catch mistakes that tools and outsiders miss. Agree in advance who makes the final call on terms both sides are unsure about.

Account access

If the service needs your search term reports, consider exporting them yourself and sharing the files rather than granting full account access. Where access is needed, use Amazon's own user permission tools so each person has their own login, never shared passwords, and remove access when the work ends.

Using the results

Start with the highest-priority groups rather than launching every term at once. That keeps budgets focused and makes early search term reports easier to read. Add the suggested negatives at the same time, and set a date to compare real search data against the research so you can see which recommendations held up.

Other marketplaces

If you sell in more than one country, ask whether the service researches each marketplace in its own language or translates a single list. Shoppers phrase searches differently from country to country, and a direct translation can produce terms nobody uses. Native-language review is worth asking about for each market in scope.

Asking for evidence of method

Rather than asking for results from other clients, which you cannot verify, ask for a short sample: how the service would treat five terms you supply, including one you suspect is irrelevant. The reasoning it gives tells you more about its filtering than any claim about past performance.