How the skill works: your agent reads query and page pairs from Google Search Console; search console property from What you tell it, 1st of the month at 9:00, and gives you a question for you and a telegram ping and the report in the chat.
Reads
Google Search ConsoleQuery and page pairs
What you tell itSearch Console property
reads
Your agentClaude Code, Codex, Cursor or Gemini1st of the month at 9:00
posts
You get
A question for youOn your board, answered with a tap
A Telegram pingOnly when it needs your answer, if you connect Telegram
The report in the chatEvery run, with or without Briefkin
What you get
Split queries: query, competing pages, clicks and position for each.
Flip-flopping: queries where Google swaps the ranking page from day to day.
Suggested owner for each query, with the reason.
The ask: which page to keep, and whether to draft the fix.
The card on the right is a sample. Yours shows your own numbers and updates in place every run.
What lands on your board
Your board1st of the month at 9:00
example.comSearch Console keyword cannibalization report · Claude Code
6 queries split clicks between two of your pages
Which page should own "invoice template"? Pick one and I will draft the internal links and the merge or redirect for the other.
6 queries with 50+ clicks are split across two pages. For "invoice template", Google swapped the ranking page on 9 of the last 28 days.
Query
Page
Clicks
Position
invoice template
/templates/invoice
610
4.2
invoice template
/blog/free-invoice-templates
480
5.8
late payment reminder
/blog/late-payment-reminder
220
3.9
late payment reminder
/templates/payment-reminder
170
6.1
net 30 meaning
/blog/net-30-payment-terms
140
4.4
net 30 meaning
/glossary/net-30
95
7.3
An example with made-up numbers. Your agent posts your real ones.
What it does
Keyword cannibalization happens slowly: a blog post, a template page and a glossary entry all creep toward the same query, Google alternates between them, and none of them ranks as well as one strong page would. Search Console shows it only if you filter one query at a time and open the Pages tab.
This skill pulls every query and page pair for the last 28 days, finds the queries where clicks are genuinely split, and checks whether Google keeps swapping the ranking URL. Then it asks you which page should own each query, because that call depends on intent and business value, not just numbers.
How it works
1Pull query and page pairs
searchanalytics.query for the last 28 days of final data with dimensions: ["query", "page"], type: "web" and rowLimit: 25000, paging with startRow until a page comes back short.
2
2Find split queries
Group rows by query. Keep queries with at least 50 clicks where two or more pages each have 15% or more of the clicks. Leave out branded queries (your brand regular expression) and URLs that differ only by a fragment or tracking parameter.
3Check for flip-flopping
For the top 10 split queries, query dimensions: ["date", "page"] with a dimensionFilterGroups filter query equals <query>. If the page with the most impressions changes from day to day, Google is undecided; flag those first.
4
4Ask who should own each query
Post an action card with each query, the competing pages, their clicks and average position, and a suggested owner (usually the page with the better position and the intent match). Offer to draft the fix: internal links, a merge, or a redirect.
Post it to the board
With BRIEFKIN_TOKEN set, the agent sends the result as one card that updates in place every run. Without it, you get the report in the chat.
Install and run it
Pick your agent, where the data comes from and how to schedule it; the commands below follow.
Agent
Data
Schedule
Install the skill
One command. It puts the SKILL.md in ~/.claude/skills and in ~/.agents/skills, which Codex, Cursor, Gemini CLI, Copilot CLI and OpenCode read too.
shellcurl -fsSL https://briefkin.dev/skills/google-search-console-keyword-cannibalization/install | sh
Run it
Ask Claude Code:
promptRun the google-search-console-keyword-cannibalization skill
Or paste the prompt instead of installing:
promptUsing the Search Console API (searchanalytics.query), pull the last 28 days of web search data with dimensions query and page, paging with startRow. Find queries with at least 50 clicks where two or more of my pages each get 15% or more of the clicks, leaving out branded queries and URLs that differ only by a fragment or parameter. For the top 10, check daily data (dimensions date and page, filtered to the query) to see whether Google keeps swapping the ranking page. Show each split query with its pages, clicks and average position, suggest which page should own it and why, and ask me which one to keep.
Put it on a schedule
1st of the month at 9:00, from cron on any machine where Claude Code is signed in:
crontab0 9 1 * * claude -p "Run the google-search-console-keyword-cannibalization skill"
Add it with crontab -e. A laptop asleep at that time runs it when it wakes if you use launchd or Task Scheduler instead; GitHub Actions works for the API route.
Send it to your board
Connect Briefkin once with curl -fsSL https://briefkin.dev/install | sh and every run lands on your board as a question for you. How setup works
SKILL.md---
name: google-search-console-keyword-cannibalization
description: "On the 1st: the Search Console queries where two or more of your pages split the clicks, with a question about which page should own each. Use when the user asks for this, or when it runs on a schedule."
---
# Search Console keyword cannibalization report
On the 1st: the Search Console queries where two or more of your pages split the clicks, with a question about which page should own each.
Keyword cannibalization happens slowly: a blog post, a template page and a glossary entry all creep toward the same query, Google alternates between them, and none of them ranks as well as one strong page would. Search Console shows it only if you filter one query at a time and open the Pages tab.
This skill pulls every query and page pair for the last 28 days, finds the queries where clicks are genuinely split, and checks whether Google keeps swapping the ranking URL. Then it asks you which page should own each query, because that call depends on intent and business value, not just numbers.
## Two ways to get the data
Use the API or MCP route when you have an MCP server, a CLI or an API token for the app. Otherwise, if you can drive a browser the user is signed in to (Claude in Chrome, a Playwright or Chrome DevTools MCP server), use the browser route. Say which one you used.
## Route A: API or MCP server
### What you need
- Read access to the Search Console property.
- A Search Console MCP server, or a service account added to the property with `GOOGLE_APPLICATION_CREDENTIALS` set and the `webmasters.readonly` scope.
If something is missing, say what and stop; do not invent numbers.
### Steps
1. **Pull query and page pairs** (Google Search Console): `searchanalytics.query` for the last 28 days of final data with `dimensions: ["query", "page"]`, `type: "web"` and `rowLimit: 25000`, paging with `startRow` until a page comes back short.
2. **Find split queries**: Group rows by query. Keep queries with at least 50 clicks where two or more pages each have 15% or more of the clicks. Leave out branded queries (your brand regular expression) and URLs that differ only by a fragment or tracking parameter.
3. **Check for flip-flopping** (Google Search Console): For the top 10 split queries, query `dimensions: ["date", "page"]` with a `dimensionFilterGroups` filter `query equals <query>`. If the page with the most impressions changes from day to day, Google is undecided; flag those first.
4. **Ask who should own each query**: Post an action card with each query, the competing pages, their clicks and average position, and a suggested owner (usually the page with the better position and the intent match). Offer to draft the fix: internal links, a merge, or a redirect.
## Route B: the browser
### What you need
- Signed in to Google with at least Restricted access to the Search Console property.
If something is missing, say what and stop; do not invent numbers.
### Steps
Start at https://search.google.com/search-console/performance/search-analytics. The user is signed in already; never type passwords or codes, and ask if a sign-in page appears.
1. **Open the Performance report for the property**: The URL opens the last property you used; switch with the property picker at the top left. Set **Date: Last 28 days** and **Search type: Web**, and make sure Clicks, Impressions and Average position are all toggled on above the chart.
2. **List the candidate queries**: Open the **Queries** tab, sort by Clicks, and keep queries with at least 50 clicks that are not your brand. The table shows 1,000 rows at most; **Export** to Google Sheets or CSV if you need more.
3. **Check each query's pages**: Click a query (this adds a query filter), then open the **Pages** tab. Keep the query when two or more pages each have 15% or more of its clicks, ignoring URLs that differ only by a fragment or a tracking parameter. Note clicks and average position for each page.
4. **Look for flip-flopping**: With the query filter on, click the first competing page to add a page filter and read its daily clicks and impressions on the **Dates** tab; remove it and repeat for the second page. When the page with more impressions changes from day to day, flag the query first.
5. **Write up the ask**: For each split query: the pages, clicks and positions, the suggested owner (the better position and intent match), and whether Google keeps swapping. Ask which page should own each one and offer to draft internal links, a merge or a redirect.
Note: Search Console never shows query and page pairs in one table, so the browser route checks one query at a time. That is fine for the top 20 to 30 queries; for a full pass over every query, use the API route or the Looker Studio connector. The last 2 to 3 days of the range are partial.
## The report
- **Split queries**: query, competing pages, clicks and position for each.
- **Flip-flopping**: queries where Google swaps the ranking page from day to day.
- **Suggested owner** for each query, with the reason.
- The ask: which page to keep, and whether to draft the fix.
## Post it to Briefkin
When `BRIEFKIN_TOKEN` is set, or `~/.config/briefkin/env` exists (the installer from https://briefkin.dev/docs writes it), also send the result to the user's Briefkin board as one card. Keep the `item_key` the same on every run so the card updates in place, use a new `event_id` per run, and put today's real numbers in place of the example ones. If the request fails, say so once and carry on.
```bash
[ -n "${BRIEFKIN_TOKEN:-}" ] || . ~/.config/briefkin/env
curl -sS -X POST "${BRIEFKIN_URL:-https://briefkin.dev}/v1/events" \
-H "authorization: Bearer $BRIEFKIN_TOKEN" \
-H "content-type: application/json" \
-d @- <<'JSON'
{
"v": 1,
"event_id": "google-search-console-keyword-cannibalization/<date of this run, e.g. 2026-10-16>",
"item_key": "google-search-console-keyword-cannibalization",
"project": "example.com",
"source": "Search Console keyword cannibalization report skill",
"kind": "action",
"title": "6 queries split clicks between two of your pages",
"summary": "**6 queries** with 50+ clicks are split across two pages. For **\"invoice template\"**, Google swapped the ranking page on 9 of the last 28 days.",
"action": {
"request": "Which page should own \"invoice template\"? Pick one and I will draft the internal links and the merge or redirect for the other.",
"options": [
"Keep /templates/invoice",
"Keep the blog post",
"Show all 6 first",
"Not a problem"
]
},
"widget": {
"type": "Table",
"props": {
"columns": [
"Query",
"Page",
"Clicks",
"Position"
],
"align": [
"left",
"left",
"right",
"right"
],
"rows": [
[
"invoice template",
"/templates/invoice",
610,
4.2
],
[
"invoice template",
"/blog/free-invoice-templates",
480,
5.8
],
[
"late payment reminder",
"/blog/late-payment-reminder",
220,
3.9
],
[
"late payment reminder",
"/templates/payment-reminder",
170,
6.1
],
[
"net 30 meaning",
"/blog/net-30-payment-terms",
140,
4.4
],
[
"net 30 meaning",
"/glossary/net-30",
95,
7.3
]
]
}
}
}
JSON
```
Widgets come from the Briefkin catalog (https://briefkin.dev/v1/catalog). Without a token, just give the user the report.
## Run it on a schedule
Suggested: 1st of the month at 9:00 (`0 9 1 * *`). From cron, with the agent signed in on that machine:
```bash
0 9 1 * * claude -p "Run the google-search-console-keyword-cannibalization skill"
# Codex: codex exec, Gemini CLI: gemini -p, Copilot CLI: copilot -p, OpenCode: opencode run
```
Or use the agent's own scheduler (Claude Code routines, Codex automations, Claude or ChatGPT scheduled tasks): https://briefkin.dev/skills/google-search-console-keyword-cannibalization#schedule
Source: https://briefkin.dev/skills/google-search-console-keyword-cannibalization
What your agent needs
Read access to the Search Console property.
A Search Console MCP server, or a service account added to the property with GOOGLE_APPLICATION_CREDENTIALS set and the webmasters.readonly scope.
Connecting Google Search Console
MCP server. Google doesn't publish one for Search Console. Community MCP servers exist; check which scopes one asks for before you sign in.
API with a service account. Enable the Google Search Console API in a Google Cloud project, create a service account, and add its email under Settings > Users and permissions on the property. Point GOOGLE_APPLICATION_CREDENTIALS at the key and set GSC_SITE_URL to sc-domain:example.com or https://www.example.com/.
OAuth as yourself. When you can't add users to the property, a desktop OAuth client with the webmasters.readonly scope and a stored refresh token works for scheduled runs.
Ask for webmasters.readonly and add the service account as a Restricted user; the reports only read.
Questions
Is every split a problem?
No. A query with mixed intent can deserve two results, for example a template page and a guide, and Google sometimes shows both at once. The skill asks rather than acting because the fix depends on intent. A split is worth fixing when the pages flip-flop or both rank below where one strong page would.
Why 28 days?
Long enough to see Google swap pages, short enough to reflect the site as it is now. Search Console keeps 16 months, so for a seasonal site you can run it over the same month last year as well.
Do I need Briefkin to use the Search Console keyword cannibalization report skill?
No. The skill runs with your own agent and gives you the report in the chat. With a Briefkin token set, each run also posts it to your board as a asks you, so you can glance at it instead of asking again.
Which agents can run it?
Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot CLI, OpenCode, Windsurf and Amp all read SKILL.md files. The install command puts it in ~/.claude/skills and ~/.agents/skills, where they look. ChatGPT, the Claude app and Grok take the prompt instead. Pick your agent under "Install and run it" and the commands change to match.
Can it work without an API or MCP server, from my browser?
Yes. Switch the data source to "Your browser" and the steps change to the Google Search Console pages themselves: the agent opens search.google.com/search-console/performance/search-analytics in a browser you are signed in to and reads the numbers off the page. Claude Code, Codex, Cursor, Claude app and ChatGPT can drive a browser out of the box; Gemini CLI, Copilot CLI and OpenCode need a Playwright or Chrome DevTools MCP server first. It is slower than the API and needs you signed in, so it suits a report you run by hand or from a scheduler on your own computer.
Answer your agents from one board, or from Telegram
Skills are free to use with any agent. Briefkin is where their results land: one board for every report, number and question your agents send, with a ping when something needs you.