Content debt is the pile of thin, outdated, and self-competing pages that quietly drag on a site's findability, and a replatform is the one moment you can actually get rid of it for free. Migrate those pages as-is and you pay to move them, pay to redirect them, and pay again when they keep splitting rankings on the new platform. Audit and cut them before cutover, and the new site launches lighter, faster to crawl, and easier for both Google and AI answer engines to read.
Now picture the last time you moved house.
You stood in the kitchen at 11pm holding three nearly identical spatulas, a lamp with a frayed cord you'd been meaning to fix since 2019, and a sealed box labelled "MISC" in your own handwriting that you were genuinely afraid to open. And what did you do? You taped them all up and paid a stranger to carry them into your nice new apartment, where they now live in the second bedroom you call an office.
That is a content migration. Every page is a box. And most of us pack the whole house.
The three flavours of stuff you're about to pay to move
Content debt isn't one problem, it's three, and they cost you in different ways. The frayed lamp is your thin content: pages that were never really finished, the auto-generated tag archives, the "coming soon" that came and went, the 90-word service stub nobody expanded. They add crawl surface and add nothing a reader or a machine can use.
The three spatulas are your cannibalizing pages: multiple URLs all reaching for the same query, so Google can't decide which one to rank and hedges by ranking none of them well. You didn't mean to make three. They accumulated. (One from the 2021 campaign, one the freelancer wrote, one the product team insisted on.)
And the sealed MISC box is your stale content: pages that were accurate once, still pull a little traffic, and are now quietly wrong. Old pricing. A discontinued feature. A statistic from a report that's been superseded twice. Terrifying to open, so nobody does.
| Debt type | What it costs | Usual verdict |
|---|---|---|
| Thin | Crawl budget, weak signals, dead-end sessions | Kill or merge |
| Cannibalizing | Split authority, unstable rankings | Consolidate into one |
| Stale but trafficked | Trust erosion, wrong answers cited by AI | Rewrite, don't cut |
Why the migration is the honest moment
Here's the thing nobody tells you: content debt is invisible until you have to inventory it. On a running site, a bad page just sits there earning its 11 visits a month, and no one's job is to notice. But a replatform forces you to build a full list of every URL and decide, one at a time, what happens to each. That list is the first honest look most teams have had at their own site in years.
So the temptation is to lift-and-shift everything. It feels safe. It is not safe. It is the moving-truck fallacy: because carrying the box feels cheaper than the emotional labour of deciding, you carry the box. Except in a migration, carrying it isn't free. Every URL you keep needs a redirect mapped, tested, and monitored. Every thin page you preserve dilutes the crawl budget on the new domain right when Google is re-evaluating everything. Every stale page you migrate becomes a page an AI answer engine might quote back to your customers with last year's price.
You don't clean the new apartment. You clean before you pack.
How to actually run the audit
Pull three signals for every page and let them argue with each other. First, organic traffic and impressions over the last twelve months, so you're not deleting something that quietly earns. Second, backlinks, because a thin page with real inbound links gets redirected, not killed. Third, query overlap between pages, which is how you catch the spatulas.
Then sort every URL into four outcomes:
- Migrate — earns traffic or links, still accurate, no better duplicate. Move it and map the redirect.
- Merge — one of several pages chasing the same intent. Pick the strongest, fold the rest in, 301 the losers into the winner.
- Rewrite — valuable URL, wrong or dated content. Keep the address, replace the contents before or right after cutover.
- Kill — no traffic, no links, no unique reason to exist. Let it 410, or redirect to the nearest relevant parent.
The one rule I'd tattoo on a project manager: a page with zero traffic and zero links is not automatically a kill. Check whether it ranks for anything, whether it's the only page covering a real topic, whether sales sends the link in emails. Traffic is a lagging indicator. Some pages are thin because they're neglected, not because they're worthless.
This is exactly the sort of scoring our crawler does before a cutover — reading duplication, orphaning, redirect risk, and staleness into a per-page migrate, rewrite, or kill call so you're not making 4,000 of these judgements by feel at 11pm. You can see how the plan comes together on our migration overview.
What I'd actually do
I'd cut hard, and I'd cut before I picked a single field mapping. My honest position: the average site heading into a replatform can retire somewhere between a fifth and a third of its URLs and come out ranking better, because the pages that remain stop competing with their own ghosts. If your inventory says otherwise, prove it with traffic and links, page by page. "But someone might need it" is not proof. It's the frayed lamp talking.
The migration will make you open every box you own exactly once. Don't tape the spatulas back up and carry them into the nice new place. You'll only be standing here again in three years, holding four of them.