You score a page as migrate, rewrite, or kill before touching a template by answering three questions on paper: does it earn traffic or links worth keeping, is the content still true and useful, and will anything break if the URL moves? Get those three right in a spreadsheet and the build becomes a chore instead of a gamble. Everything else is decoration.
Picture the closet. You know the one. You open it to grab a jacket and six other things slump out at your ankles, and standing there holding a sweater you last wore during a different administration, you realize you own forty shirts and wear the same four. A CMS migration is that closet, except the closet is your whole site and someone has set a date for the movers.
Here is the mistake almost everyone makes. They rent the new place first (spin up the new platform), start carrying boxes over (migrating pages), and only sort the junk once it is already stacked in the new living room. Now you are paying new-place rent to store a sweater from 2015. Score first. Carry second.
The three piles
Sorting a closet, you end up with three heaps on the bed whether you plan to or not. The keep pile. The needs-tailoring pile. And the pile you cannot believe you held onto this long.
Pages work exactly the same way.
| Pile | Verdict | What it means | The test |
|---|---|---|---|
| Keep | Migrate | Still fits, still earns its hanger space | Gets traffic or links, content is accurate, URL and references are clean |
| Tailor | Rewrite | Good bones, bad shape | Topic still matters but the content is stale, thin, duplicated, or dated |
| Bin | Kill | You are keeping it out of guilt | No traffic, no links, no business reason, nobody will miss it |
Three columns. That is the whole product, honestly. The hard part is being honest about which pile a page belongs in, because every page feels like the jacket you might wear someday.
What actually goes on the scorecard
A page earns its spot in the keep pile with evidence, not affection. Pull organic traffic over the last twelve months, not last month, because seasonality lies. Pull inbound links, because a page can get almost no traffic and still be the thing forty other sites point at (kill that and you have thrown away equity you cannot rebuild). Check whether it ranks for anything you care about. Note whether anyone internally will fight to the death for it, because the CEO's favorite thought-leadership post is a political fact whether or not the data agrees.
Most sites find the same thing when they run this honestly: a small slice of URLs carries the overwhelming majority of the value, and a long, dusty tail does nothing but age. That tail is not neutral. It costs you crawl budget, it dilutes your topical focus, and it gives both Google and the AI answer engines more low-quality pages to misread. A page that helps nobody is not free just because it is quiet.
The tailoring pile is where judgment lives. These are pages on topics that still matter where the content has gone soft: a pricing page listing numbers from two redesigns ago, three near-identical articles that should be one, a guide that was accurate before the product changed. You are not throwing out the garment, you are taking it in. Mark what the rewrite needs so it is a brief, not a mystery, when someone picks it up.
And the bin. Expired campaigns, duplicate PDFs, that microsite from the trade show, the tag archive with one post in it. Killing a page does not mean the URL evaporates. If it had any links or traffic at all, it earns a redirect to its nearest living relative. Google treats a 301 as passing ranking signals, so the right move for a dead page is often a clean forward, not a 404. Killing the content and keeping the exit sign are two different decisions, and you make both at the scoring stage.
Why on paper, and why now
Because once a page is in the new system it stops being a choice and starts being a sunk cost. Someone built the template, someone mapped the fields, someone QA'd it. Asking "should this exist?" after all that feels rude. So nobody asks, and the junk rides along, and you have faithfully reproduced your old mess on an expensive new platform.
Scoring on paper is cheap, reversible, and it happens before anyone is emotionally invested in a page surviving. A spreadsheet does not care about your feelings. That is its entire charm.
This is the slice of the work that tooling like ours is built for: crawl everything, pull traffic and links and duplication and orphan signals against each URL, and hand you a scored list so the migrate/rewrite/kill call is a five-second read per row instead of an afternoon of detective work. The judgment stays yours. The evidence arrives pre-sorted.
So what would I actually do
Score the whole site before the new platform exists, and treat the kill pile as the win, not the loss. I would rather cut a quarter of a site and migrate something lean, accurate, and findable than lovingly rebuild every orphan and call completeness a strategy. The projects that go badly are the ones that migrated everything "to be safe." Safe is how you end up paying new-apartment rent to store the sweater.
Open the closet. Make the three piles. The jacket you might wear someday is the one you will regret carrying up the stairs.