Your duplicates aren't three chances to rank

Search your own help center for something basic. "Invite a user", say.

If you get 3 results (an old getting-started article, a newer feature page, and something a support lead wrote in a hurry during onboarding season), you don't have 3 chances to be found. You have one chance split 3 ways, and a bot that will answer differently depending on which fragment it happens to pick up.

Why people end up here

Nobody sets out to write the same article 3 times. It happens because writing a new page is easier than finding and editing the old one, because the product team asked for "a page about the new flow", and because nobody owns the question of what already exists.

It also used to be survivable. A human looking at 3 similar results picks the most recent-looking one and gets on with their day.

What duplicates do to retrieval

A retrieval system doesn't pick the best one. It scores passages by similarity to the question, and near-identical passages score near-identically. So which one wins is close to arbitrary, and can change between two customers asking the same thing an hour apart.

Worse: your three versions almost certainly disagree. Not dramatically: one mentions the seat limit, one doesn't; one was written before the permissions changed. Those small disagreements are exactly what turns into a confident wrong answer.

Merge, don't tidy

The instinct is to improve all 3. Don't. Pick the one with the most traffic and the best URL, fold everything true from the others into it, and redirect the rest to it.

Redirect, don't delete. You keep the search equity you built up, and anyone with an old link still lands somewhere useful.

One article that fully answers the question beats 3 that partly answer it, and it's less to keep true next quarter. That last bit is the real prize.

Finding them

Sort your articles by title and read the list. Duplicates are usually obvious once they're next to each other, which they never are in a category tree. Or run check 4 of the free health check on your busiest topic to count how many articles cover it.