Picture the driveway. The truck is rented, the movers are on the clock, and you're standing in the garage holding a box labeled "MISC \u2014 kitchen?" that you last opened during a different presidency. This is the exact moment nobody wants to make a decision, and it's the exact moment the decision costs the most. So here's the whole trick to a content migration: you decide whether a page moves, gets rewritten, or dies before you build a single template, by scoring it against three plain questions. Does anyone reach it? Does it still tell the truth? Would you miss it if it quietly vanished?
I keep asking teams to do this on paper first because I keep watching them do the opposite. Somebody rebuilds four thousand pages in a shiny new platform, launches, and then discovers that more than half of them hadn't drawn a human visitor since the last redesign. They paid full freight to move boxes of junk into a nicer house. Which, if you've ever moved, you already know is the whole game: the packing is where you win or lose, not the unloading.
Score in the old house, not the new one
The mistake is treating scoring as a launch task. It isn't. Every hour you spend fighting a template, mapping a field, or babysitting a redirect for a page that should have died is an hour billed at moving-day rates. Sorting the box costs almost nothing while it's still sitting on your old garage floor. Sorting it after the movers have carried it up two flights of stairs costs you the stairs, twice.
So before anyone touches a theme, an export, or a single content model, you walk the whole house with a marker. Every page gets one of three marks. Keep (migrate as-is). Refurbish (worth having, but not in this condition). Curb (goodbye).
What actually tips a page into each pile
You don't need a philosophy degree for this. You need a few signals you can pull from analytics, your log files, and a link tool, then a rule for what they mean together. Here's the shape of it.
| Signal | Leans migrate | Leans rewrite | Leans kill |
|---|---|---|---|
| Organic entrances (last 12 months) | Steady traffic | Some, but declining | Zero or near-zero |
| External links pointing in | Earned links you'd hate to lose | A few, from thin sources | None |
| Accuracy / freshness | Still true today | Right topic, stale facts | About a product you sunset |
| Duplication | Unique | Overlaps a stronger page | Near-identical twin exists |
| Findability | Linked in nav or hub | Buried but reachable | Orphaned, no internal links |
The point of a table like this is to force honesty. A page with zero entrances, zero external links, no internal links, and a near-identical twin isn't a hard call. It's a box of expired condiments. You are allowed to throw it out. In fact, you are obligated to, because every dead page you carry over dilutes the crawl budget and the topical clarity of the ones that actually earn their keep.
Migrate, rewrite, kill \u2014 and what each one promises
Migrate means you move the URL and its content essentially intact, and you owe it a clean 301 to the same path (or a mapped new one). A 301 passes ranking signals along, which is exactly why you protect the pages that have signals worth passing. Don't waste that on junk.
Rewrite is the pile people abuse. It's not "migrate later." It's a page whose topic still earns a spot but whose body has to be redone before it goes up. The trap is marking everything "rewrite" to avoid an argument, then discovering at launch that you've committed to writing six hundred pages nobody scoped. If you can't name who's writing it and when, it isn't rewrite. It's kill wearing a disguise.
Kill means the URL retires. Sometimes that's a 301 to the nearest better page (a donation to a good home). Sometimes it's a clean 410, telling crawlers to stop asking. What kill is not is a silent 404 you discover in Search Console three weeks after cutover, when the box you left on the curb turns out to have had your passport in it.
Why paper beats the sandbox
Here's the part that saves you money. A spreadsheet where every URL has a score, a reason, and a destination is a plan you can argue about, hand to a developer, and check off. A half-built staging site is a place where you learn the same facts, later, after you've paid to build the thing you're now deleting. This is precisely the sort of scoring a crawler is good at doing across tens of thousands of URLs at once \u2014 pulling traffic, links, duplication, and orphan status into one verdict per page \u2014 so the humans spend their judgment on the genuinely close calls instead of the obvious condiment boxes. If you want the mapping side of it, the migration guides get into the redirect mechanics.
Would I ever skip the scoring and just move everything "to be safe"? No. "Move everything" is not safe. It's the most expensive form of indecision there is, because you pay the moving cost now and the clutter cost forever.
So do the boring thing. Walk the whole house with the marker before the truck shows up. Mark the box, commit to the mark, and when you find the one labeled "MISC \u2014 kitchen?", have the nerve to open it. Ninety percent of the time it's a fondue set you've never used. The other ten percent, it's the passport. Scoring is just knowing which is which before the truck pulls away.