You score a page migrate, rewrite, or kill before you touch a template by grading it on evidence you already have: its traffic, its inbound links, whether anything actually points at it, whether a near-identical page exists three URLs over, and whether a human or a machine could find a reason for it to exist. Migrate the pages that earn their keep as-is. Rewrite the ones with real demand and rotten execution. Kill the rest, on paper, before they get a seat in the new site.
Here is the mistake I keep watching people make, and it is a very human one.
Imagine you are planning a wedding, and instead of writing the guest list first, you book the biggest ballroom in town "to be safe." Then you send an invite to everyone. Your dentist. That cousin. The guy from the gym whose name you are 80% sure of. You pay per head, you seat all of them, and only at the reception do you realize half the room has no idea why they're there. That is a content migration with no triage. You rebuild templates, you wire up redirects, you QA every field, and months later you find out a third of what you carried over hadn't drawn a visitor since the last administration.
The three verdicts, in plain terms
The whole exercise collapses into three decisions, and you want to make each one with the page's file open and its stats next to it, not with a half-built CMS in front of you.
Migrate means the page works and you move it more or less intact. It has an audience, it has links, it answers something. You are inviting it, and it already knows how to behave at a party.
Rewrite means the topic deserves a spot but the current page does not. There's demand, maybe old rankings, maybe a valuable URL other sites still link to, but the content is thin, dated, or written for a version of the business that no longer exists. Invited, on the condition it shows up in a clean shirt.
Kill means it doesn't make the cut. No traffic, no links, no distinct purpose, or three other pages saying the same thing slightly worse. You either let the URL 404 (with intent) or, more often, redirect it to the nearest page that did get invited. Killing isn't cruelty. It's respect for the people who have to maintain the room afterward.
The signals you can score on paper
You don't need the new platform to answer any of this. Every signal below lives in analytics, a log file, a crawl, or Search Console. Pull them into one row per URL and the verdict mostly writes itself.
| Signal | Where it lives | What it tells you |
|---|---|---|
| Organic sessions, trailing 12 months | Analytics | Whether anyone arrives at all |
| Referring domains | Backlink tool | Whether killing it snaps a link someone else owns |
| Impressions and ranking queries | Search Console | Latent demand a rewrite could capture |
| Internal links in | Site crawl | Whether it's orphaned or load-bearing |
| Content similarity | Crawl / dedupe pass | Which near-duplicate is the keeper |
| Last meaningful update | CMS export | Whether the facts have expired |
Score those six and you can defend nearly any verdict to a nervous stakeholder, which you will need to, because someone always loves the page nobody visits. (There is always one.)
Where to draw the lines
The thresholds are yours to set, but commit to them in writing before you look at a single URL, or you'll bend them for every page with a sad face attached. A page with real traffic and healthy links: migrate, and don't overthink it. A page with impressions or referring domains but no traffic and a stale body: rewrite, because the demand is proven and only the execution failed. A page with zero organic sessions over a year, zero referring domains, no internal links pointing in, and a twin elsewhere on the site: kill it, and redirect the URL to its twin so the twin inherits any residual equity.
The awkward middle is the whole game. That's a page with a trickle of traffic and no links, or a link but no traffic. My rule: if a page has demand you can measure, it's a rewrite, never a kill. Killing measured demand is how you wake up to a traffic cliff three weeks after cutover and spend a Saturday reverse-engineering which URL you dropped.
Why the paper has to come before the template
Because everything gets more expensive once you're building. On a spreadsheet, a kill is a keystroke and a redirect rule. In a half-migrated CMS, that same page has a template, a content type, mapped fields, a QA ticket, and an emotional attachment from whoever spent an afternoon on it. You end up migrating things purely because you already migrated them. Sunk cost, but for web pages.
Scoring first also shrinks the actual build. If a clean-eyed pass tells you a meaningful chunk of the site is kill-or-merge, you're suddenly modeling templates for the pages that matter instead of every ghost in the archive. Cheaper, faster, and the new site launches lighter than the old one instead of hauling the debt across. This is the entire job a tool like ours does before you migrate: crawl, score, and hand you a defensible migrate, rewrite, or kill plan while it's still just rows in a sheet.
So do I actually believe you can decide this on paper, before a single template exists? Yes, completely, and I'll go further: if you can't make the call from the data, that's not a reason to defer it to the sandbox, it's a sign the page never had a reason to exist. Write the guest list at the kitchen table. Book the venue for the people who are actually coming. And when your dentist asks why he wasn't invited to the new site, you'll have the spreadsheet, and the spreadsheet does not feel guilt.