Home / SEO / KA-015

Cluster a keyword export into topic groups and map each to one page

Turn a raw keyword export into topic groups, give each group one target page, and flag groups where two of your pages compete.

Who it suits

Freelance SEO, small agency owner

Time saved

About 2 to 3 hours per 1,000 keyword export (estimate)

Level

Medium

Updated

Oct 13, 2026

Why this task

Keyword exports arrive as long flat lists, so teams either ignore them or build one page per phrase. That spreads effort thin and sets your own pages against each other. An agent can group the list by topic, tie each group to one page, and show where two pages already share a group. You then spend your time on the judgment calls.

You are helping me organise keyword data for [SITE NAME], a [BUSINESS TYPE] serving [LOCATION OR MARKET].

Below are two inputs.

Input 1, keyword rows from [SOURCE: Google Search Console or keyword tool]. Columns: [COLUMN NAMES].
[PASTE KEYWORD ROWS]

Input 2, my current pages, one per line as URL | title | purpose.
[PASTE PAGE LIST]

Do the following.

1. Group the keywords into topic clusters. Put two keywords in the same cluster only if one page could fully answer both searches. Give each cluster a short plain name.
2. For each cluster, list its keywords and the total of [METRIC: impressions or volume] from my data. Do not estimate numbers I did not give you.
3. Map each cluster to exactly one page from Input 2. If no page fits, write NEW PAGE NEEDED and suggest a working title.
4. Flag cannibalization. Flag a cluster when my data shows two or more of my pages getting impressions for its keywords. Also flag it when two pages in Input 2 clearly target it. Name the pages and the evidence.
5. List keywords you could not place, with a one line reason.

Rules:
- Use only the data above. If a column is missing, say so.
- Keep cluster names stable. Earlier batches used these names: [PREVIOUS CLUSTER NAMES OR NONE].
- Mark any guess about searcher meaning as UNSURE.

Output a table with columns: cluster, keywords, total [METRIC], target page, cannibalization flag, evidence, notes. Then give the unplaced list.

Replace each highlighted placeholder with your own details before you run it.

Steps

1. Export your data. From Google Search Console, export queries with their pages for the last 3 months. From a keyword tool, export any target keywords you want to add.
2. Clean the sheet before the agent sees it. Remove branded queries, obvious spam queries and rows with no impressions or volume. Keep one column per field.
3. Write a short list of your current pages with their web address (URL), title and main purpose. The agent needs this to map groups to real pages.
4. Paste the cleaned rows and the page list into the prompt below. For large files, send them in batches of a few hundred rows and keep the cluster names stable between batches.
5. Read the cluster list first. Merge groups that a searcher would treat as the same question, and split groups that mix two different needs.
6. Check every cannibalization flag in Google Search Console. Filter by the query and look at which pages earn impressions, and whether the ranking page changes over time.
7. Decide one action per flagged group: merge the pages, redirect one, change one page's focus, or leave both if they serve different needs.
8. Save the final page map in your spreadsheet with an owner and a review date. Repeat the exercise each quarter with a fresh export.

What you get

A table where every keyword sits in one named cluster and every cluster points to one page or a clear new page suggestion. Cannibalization flags name both pages and the rows that show the overlap. Unplaced keywords are listed with reasons, so nothing vanishes quietly.

What can go wrong, and the check for each

Related playbooks