Scoring a page before a migration means deciding, on evidence you already have, whether it earns a seat on the new site (migrate), needs fixing before it does (rewrite), or should be left behind with a redirect (kill). You do all of it in a spreadsheet, weeks before a single template exists. No sandbox, no staging URL, no build. Just a row per page and a verdict.
Here is the moment that provokes this, every single time. You open the export. The crawl comes back with a number you were not emotionally prepared for. Fourteen thousand URLs. Maybe forty. And somebody in the kickoff meeting says the words that have doomed more migrations than any plugin ever did: "Let's just move everything and clean it up later."
You will not clean it up later. You never have. Nobody ever has.
Think of it as a report card, not a packing list
A packing list asks one question about each thing: does it come or not. That is how you end up moving a filing cabinet full of 2014 press releases into a brand-new house because, well, it was already there and the truck had room.
A report card is harsher and kinder at the same time. It grades each page on how it actually performed, and it hands out exactly three marks. Honor roll pages move as-is. The ones who are bright but turned in a sloppy paper go to summer school before they're allowed up. And the ones who stopped showing up months ago, contribute nothing, and are quietly dragging down the whole average? They don't get held back. They get a forwarding address and a fond farewell.
Every page reapplies. That's the whole philosophy. Legacy is not a qualification.
The three grades, and what earns each one
| Grade | What it means | Typical signals |
|---|---|---|
| Migrate | Moves roughly as-is. It earns its URL. | Steady organic traffic, inbound links, conversions or assisted conversions, cited by AI answer engines, unique content, clean metadata. |
| Rewrite | The topic deserves a spot; this version doesn't. Fix before cutover. | Thin or duplicated body, outdated facts, decent impressions but awful click-through, cannibalizing a sibling page, good links pointing at bad content. |
| Kill | Leave it behind. Redirect to the nearest living relative. | Zero traffic for 12+ months, no inbound links, orphaned (nothing on the site links to it), expired campaign, exact duplicate of a better page. |
The grade isn't a gut call. It's the subjects stacked up next to each other. A page with no traffic and no links and no citations is an easy expulsion. A page with no traffic but twelve referring domains is not a kill, it's a mystery you need to solve before you touch it.
The subjects you grade on
Four report-card subjects carry most of the decision. Organic performance over a trailing twelve months, so one dead August doesn't flunk a seasonal page. Inbound links, because a redirect that drops link equity is you setting money on fire to save a spreadsheet row. Internal links, which tell you whether the page is part of the building or a room with no door (orphans are usually kills, but check first). And whether machines can actually read and quote the thing, because an answer engine citing your page is a vote you cannot see in your analytics but absolutely want to keep.
Score those four, and the verdict mostly writes itself. The art is in the pages where the subjects disagree.
The dropout who's secretly carrying the GPA
Here's the one that ruins migrations. There is always a page that looks like a textbook kill. Ugly URL, no traffic in your analytics, last edited during a previous administration. Every instinct says cut it.
Then you check, and it's the number-one organic result for a term that quietly drives half your demo requests, or it's the single page every AI assistant quotes when someone asks what your company does. The traffic was invisible because it converts somewhere downstream, or because the answer box ate the click before it ever reached you.
Kill that page and redirect it to your homepage, and you will spend the next two quarters wondering why the pipeline went soft. (It went soft because you expelled the valedictorian for having a bad locker.)
This is the entire reason to score on paper first. You are looking for the gap between what a page looks like and what it does. A crawl that stitches your analytics, link data, and crawl structure onto one row per URL is what surfaces those gaps before they become a 301 you regret. Scoring every page this way is the core of a sane migration plan, and it is deeply, gloriously boring to do in a spreadsheet and deeply painful to discover in production.
Do it before the build, not during it
Scoring in a sandbox is scoring too late. By the time there's a staging environment, templates are built, URLs are half-decided, and every "should this even exist" conversation now costs engineering time instead of a cell edit. The spreadsheet is where a page costs nothing to kill. The staging site is where killing it starts a meeting.
So here's what I actually do, no hedging. I score the full inventory on paper first, every time, even when the client swears the site is small and tidy. I have never once found it as small or as tidy as promised. I treat "migrate everything" as a decision not to decide, and I treat every orphan and every zero-traffic page as a question rather than a verdict until the links and citations confirm it. Migrate should be the grade you have to earn, not the default everyone inherits for showing up.
Grade them on paper. Because the only thing worse than telling a page it didn't make the cut is finding out, three months after launch, that the one you forwarded to the homepage was the smartest kid in the building.