Picture the night before the movers show up. You're standing in the middle of your apartment holding three rolls of colored dots, and every single thing you own is about to force a decision: the couch goes on the truck as-is, the wobbly bookshelf gets fixed once it lands, and the treadmill you've used as a coat rack for four years goes to the curb. That's the whole job. Before you migrate a website, you score every page the same way: green means migrate it as-is, yellow means migrate but rewrite it, red means kill it and redirect. You do this in a spreadsheet, based on whether the page earns traffic or links, whether a human or a machine still needs it, and whether it's a duplicate or an orphan. No templates. No sandbox. Dots first.
The reason I'm this insistent is that I keep watching teams do it backwards. They stand up the new platform, start dragging pages across, and only then discover that four thousand of them are thin, duplicated, or pointing nowhere. Now you're paying a developer's hourly rate to move furniture you were always going to throw away.
Why the dots go on before the truck arrives
A page that shouldn't survive the move costs you three separate ways if you migrate it anyway. It costs build time to bring across. It costs crawl budget once it's live, because Google is not obligated to spend its time re-reading your 2016 press releases. And it costs you clarity, because every dead page you keep makes the good pages harder for both people and answer engines to find.
Killing a page is not free either. Every red dot needs a redirect, and Google's own guidance is to keep a 301 in place for at least a year so the signals transfer and the link equity follows. So "kill" doesn't mean "delete and walk away." It means "stop maintaining the room, but leave the forwarding address taped to the door."
What earns a green dot
Green is the couch. It goes on the truck exactly as it is because it works and everyone loves it. A page earns green when it pulls organic traffic, holds backlinks from sites you don't control, converts, or answers a question people actually ask. If a page is doing a job today, your only mission during migration is: do no harm. Same URL where possible, same content, same internal links pointing at it.
Here's the trap. Green does not mean "a stakeholder is emotionally attached to it." I have watched a VP fight to migrate a page that had eleven visits in a year, nine of which were the VP checking whether it was still there. Attachment is not traffic. Put the dot on the data.
What earns a yellow dot
Yellow is the wobbly bookshelf. Good bones, worth keeping, but you're not putting it back together in the new place without tightening the screws. These are pages that rank but read like they were written for a search engine in 2018, pages missing the structured answer that an AI engine would actually quote, pages that half-duplicate three siblings and need to be merged into one strong version. Yellow is where most of the real work lives, and it's also where you'll be tempted to lie to yourself. "We'll rewrite it later" is how a yellow dot quietly turns green and ships broken. If you can't name who rewrites it and roughly when, it isn't yellow. It's red.
What earns a red dot
Red is the treadmill. It had a purpose once. It has a coat on it now. Expired campaigns, duplicated tag pages, orphaned PDFs nobody has linked to since a reorg, near-identical location pages generated by a plugin that's no longer installed. Kill them, map a sensible redirect to the nearest living relative, and enjoy the smaller, faster site on the other side.
| Dot | Verdict | Signals that put it here | What you do at cutover |
|---|---|---|---|
| Green | Migrate | Earns traffic, holds backlinks, converts, answers a real query | Move as-is, preserve the URL, keep inbound links intact |
| Yellow | Rewrite | Ranks but thin, outdated, duplicated, or unreadable to machines | Merge or rewrite with a named owner before it ships |
| Red | Kill | No traffic, no links, orphaned, duplicate, or expired | 301 to the nearest living page, keep the redirect a year |
The part everyone underestimates: can a machine even read it?
You can score every page for traffic and links and still miss the question that decides your next two years of visibility: can Google and the answer engines actually parse this thing? A page locked behind a script, buried with no internal links, or duplicated under five URL variations is invisible no matter how brilliant the copy is. That's the layer worth automating. Crawling a whole site and flagging the duplicates, the orphans, and the redirect risks before you assign a single dot is exactly the boring, essential work a tool like ours does so you're arguing over judgment calls instead of counting PDFs by hand. Start the plan at replatformradar.com/migrate and put the dots on paper.
So what would I actually do
I default every page to red and make it earn its way up. Guilty until proven useful. It sounds harsh, and the first stakeholder review is genuinely uncomfortable, but it flips the burden of proof onto the page instead of onto you, and that's the only version of this that survives contact with a deadline. If a page can't show me a number, it doesn't get on the truck.
Score it at the kitchen table with three rolls of dots. Because the alternative is discovering the treadmill in your new living room, coat still on it, wondering how it followed you all the way here.