Cannibalization

Two of your own pages competing for the same search, and which one should win.

In appWorkspaces›Cannibalization

What the list shows

Each row is a pair of your pages that both rank for the same queries. The pair at the top is the one with the most search traffic at stake, so start there rather than working down the alphabet.

The severity band tells you how much is at stake. The counts beside it are how many queries the two pages overlap on.

Reading a pair

Open a row and you get both pages side by side: what each ranks for, which one Google prefers, and how often the winner changes. A pair where the winner keeps switching is usually worse than a pair with a clear loser, because neither page ever builds up momentum.

The recommendation names which page should win and why. It's based on which one already has the stronger signals, not on which one you'd prefer.

What to do with it

Every pair has two actions. Create brief sends the winning page to Content Studio with the conflict and the queries already filled in, so the rebuild starts from the problem rather than a blank page.

Ask Advisor puts the pair to your Strategist, which talks the merge through in plain words: whether to redirect, combine or keep both and separate them.

When there's nothing in the list

An empty list means the analyzer ran and found no pairs worth acting on. It isn't a failure, and it's common on smaller sites.

Working a finding

Open any finding and you can take it somewhere else without retyping it. Ask Advisor opens the Strategic Advisor with that finding and its full context already loaded, so you can ask why it matters or what to do first and get an answer using your own numbers. Create brief (or Rebuild in Studio) opens a Content Studio session for the page, pre-filled the same way.

Both carry the finding with them. You're not starting a conversation or a brief from nothing, which is the whole point of launching them from here rather than from their own screens.

Most findings also carry an AI Playbook: a written recommendation for that specific page. Some are generated automatically after a run, the rest are generated when you ask. You can also tick several rows and generate advice for all of them in one go, with the cost shown before anything runs. How many are written automatically is the Auto-generate for top setting in that analyzer's Configure panel, and zero means they're only written on request.

Mark applied does more than tick it off. The page joins Impact Tracking in Command Center, which then follows its clicks, impressions and position from that date onward, so you can see whether the change did anything.

Getting rid of one you don't want

Ignore on any finding hides it. It leaves the queue and moves to the Ignored pill in the filter rail, where Restore puts it back, so nothing is destroyed. The Ignored pill only appears once you've ignored something.

Ignoring handles one finding. If the same pages keep coming back, add them to Excluded pages in Settings, under Data Pipeline: find Cannibalization in the analyzer list and click Configure. One pattern per line, * as a wildcard, and plain text matches anywhere in the URL. Matching pages stop becoming findings from the next run.

That's the difference worth knowing: ignoring hides one finding, excluding a page stops future findings for it. Tag pages, author archives and paginated listings are the usual candidates.