Before you build a single template, every page on your site should have to clear customs. That means a one-line verdict per URL, decided on evidence, not affection: migrate (it's clean, wave it through), rewrite (it's carrying something you can't move as-is, repack it), or kill (it's contraband, it does not board the plane). You make these calls in a spreadsheet against real numbers, weeks before anyone opens the new CMS. That's the whole method. The rest of this is how to run the desk.
Imagine this: you're at the airport, forty minutes late, dragging three suitcases you packed in a panic. One's fine. One's nine kilos over and the zipper's crying. And one, if we're honest, contains a rock you picked up in 2014 and have carried ever since. The customs officer does not care how long you've owned the rock. Neither should you.
Why the desk exists at all
Here's the number that should end most migration debates before they start: an Ahrefs study of about a billion pages found that 90.63% of them get zero organic search traffic from Google. Not a little. Zero. Your site is almost certainly the same shape, which means most of what you're about to lovingly re-platform is a rock in a suitcase. Moving it costs you build time, QA time, redirect entries, and crawl budget on the new site. Killing it costs you the courage to look at the number and believe it.
The reason you do this on paper, not in a sandbox, is the same reason you weigh the bag at home. Discovering a page is worthless after you've templated it, migrated it, and mapped a redirect to it is the migration equivalent of repacking on the terminal floor while people step over you.
What each page has to declare
Give every URL the same short interview. What organic traffic did it pull over the last twelve months (twelve, so you catch seasonal pages that look dead in July)? How many referring domains point at it, because a page with real backlinks is a page other people vouched for, and you break those links at your peril? Does it convert, assist, or feed a funnel? Is it actually indexed right now, or has Google quietly stopped caring? And the one everyone skips: can a machine read it cleanly, or is the real content trapped in a PDF, an image, or a JavaScript widget that renders like a locked briefcase?
Those five answers do almost all the sorting for you. A page with traffic, links, and clean markup is a migrate. A page with links or traffic but a mess inside is a rewrite. A page with none of the above is a kill, and the fact that you wrote it in 2016 and are proud of it is, I'm sorry, not on the form.
The three verdicts, side by side
| Signal | Migrate | Rewrite | Kill |
|---|---|---|---|
| Organic traffic (12 mo) | Meaningful and steady | Low but has links or intent | Effectively zero |
| Referring domains | Yes, keep the equity | Some, worth preserving the URL | None |
| Business value | Converts or supports | Right topic, wrong execution | Redundant or obsolete |
| Machine-readable | Clean HTML today | Content stuck in PDF/image/JS | Doesn't matter, it's leaving |
| Action | Move as-is, map 1:1 | Rebuild the content, keep URL | 410, or 301 to the nearest living page |
One honest caveat: traffic alone lies. A compliance page or a returns policy might pull almost no organic visits and still be load-bearing for the business. That's why the interview has five questions and not one. Judgment survives; it just gets to work from the same evidence for every page instead of vibes.
The rewrite pile is where the money is
Migrate is boring and kill is satisfying, but rewrite is where a replatform actually earns its budget. That over-stuffed suitcase, the page ranking on a topic you care about with content that reads like it was faxed, is your best opportunity. You're already in there rebuilding the template. Rewriting the words while you're at it costs a fraction of what a standalone content project would, and the new page inherits the URL's history. Do not let rewrite quietly collapse into migrate because the deadline got tight. That's how debt walks across the border with you.
Kill deserves a word too. Killing a page is not deleting it into the void and hoping. Every dead URL needs a decision: a 410 if it's genuinely gone and had no value, or a 301 to the closest living relative if anything at all links to it. This is the part people defer, and deferring it is exactly what makes the second half of a migration cost more than the first.
Run it before the build, every time
Scoring thousands of URLs by hand is grim, which is the honest reason it gets skipped and everything gets migrated by default. This is the part we built a product to do: crawl the whole site, pull the traffic, link, indexation and readability signals together, and hand you a migrate/rewrite/kill column you can argue with, per URL. Argue with it. It's a starting verdict, not a sentence.
So here's what I actually think, no hedging: never migrate by default. Default-migrate is the rock in the suitcase, and it follows you to every site you'll ever build. Weigh every bag at home. The officer at the desk does not care how long you've owned the page, how hard it was to write, or that it once did numbers in 2016. It clears customs on today's evidence, or it stays on the tarmac.
