You score a page as migrate, rewrite, or kill by asking three questions before you touch a single template: does anything point to it (links, traffic, or an AI engine quoting it), is anyone actually reading it, and would you be proud to publish it today. All three yes, it flies as-is. Good bones, tired content, it goes to rewrite. Nothing points to it and nobody visits, it doesn't get a seat. Everything else is detail.
Picture the gate on a full flight. The desk calls boarding groups, scanners beep, and a handful of people stand to one side holding paper, hoping. That little drama is exactly what should happen to your pages before a replatform, and the tragedy of most migrations is that it doesn't. Someone exports the sitemap, it says 14,000 URLs, the dev quote triples, and then every single page shuffles onto the plane whether it can fly or not.
Nobody checks the boarding passes. (Then you wonder why the new site feels like a storage unit with a nicer lobby.)
Boarding pass, standby, denied
Three verdicts, three outcomes at the gate. A confirmed seat is a migrate: the page earns its keep, the content is current, and you move it roughly intact. Standby is a rewrite: the URL has value you don't want to lose, but the thing living at that URL hasn't aged well. You keep the address, you replace the passenger. Denied boarding is a kill: no links, no traffic, no reason to exist, and carrying it over just means paying to move a box you'll never open.
The point of scoring on paper, in a spreadsheet, before anyone opens a sandbox, is that it's reversible and it's cheap. Changing a cell from "migrate" to "kill" costs nothing. Un-building a template at 2am during cutover costs a weekend and your mood.
| Verdict | What it means | Signals that put it here |
|---|---|---|
| Migrate | Confirmed seat. Move as-is. | Steady organic traffic, inbound links, conversions, or AI citations. Content still accurate. |
| Rewrite | Standby. Keep the URL, replace the content. | Earns traffic or links on topic, but the page is thin, dated, duplicated, or embarrassing. |
| Kill | Denied boarding. Leave it behind. | No traffic in 12 months, no links, orphaned, superseded, or an accidental duplicate. |
The scanner reads three things
First, who points at the page. Pull inbound links and referring domains, and pull your top landing pages from analytics over the last twelve months. A page with real backlinks is almost never a kill, even if it looks sleepy, because those links are value other people built for you. Throwing it away without a redirect is like shredding a check.
Second, who reads it. Twelve months of zero organic entrances, zero internal clicks, and zero assisted conversions is a page that quietly left the building a while ago. You're just finding the empty chair now. Be honest about the window though: seasonal pages and legal or policy pages can look dead and still be load-bearing, so flag those before the scanner bins them.
Third, whether you'd publish it today. This is the one you can't fully automate, and it's where thin content, near-duplicates, and the fourteenth slightly-different "about our values" page reveal themselves. If the answer is "ugh, no," but the URL earns traffic, that's your standby pile. That's a rewrite, not a kill.
Run all three and most pages sort themselves. The confident migrates and the obvious kills fall out fast. The interesting fights are always in the standby lane, which is also where the budget conversation actually lives, because rewrites cost human hours in a way that moving a healthy page does not.
Why paper beats the sandbox
Here's the trap. Once a page exists in the new system, it has an advocate. Someone built the template, someone mapped the fields, and now killing it means admitting that work was wasted. Pages that should never have boarded get defended simply because they already took off. Scoring first, while everything is still a row in a sheet, strips out the sunk-cost feelings. A row doesn't have a champion.
This is the unglamorous half of a migration that nobody budgets for, and it's precisely what crawling and scoring a site before the move is meant to surface: the duplicates, the orphans, the pages one redirect away from a dead end, and the handful that answer engines actually cite. A tool like Replatform Radar exists to put a verdict next to every URL before the trucks arrive, but you can start with an export, analytics, and a backlink pull. The method matters more than the software.
One caution. A kill is not a delete. A killed page still needs a redirect to the nearest living relative so the mail forwards and so the links it earned flow somewhere useful. Denied boarding, then a gentle escort to a page that can actually help. There's a difference between a page that doesn't deserve a seat and one you pretend never existed.
So what would I actually do
Score everything before a single template gets touched, every time, no exceptions. I'd rather spend two days in a spreadsheet than two weeks un-migrating pages that lobbied their way onto the plane. The discipline isn't the scoring, it's refusing to build anything until the scoring is done, because the moment pages exist in the new CMS they grow advocates and the cut gets ten times harder.
If you only do one thing: make every page show its boarding pass at the gate. The ones holding nothing were never getting on anyway.