Somewhere on the site you're about to migrate is a page offering fifteen percent off a product you stopped selling three years ago, and it is still live, still crawlable, still waiting for a visitor who is never coming. Nobody put it there on purpose. It just never got deleted, the way nothing on a big site ever quite gets deleted. And here's what I keep watching teams do: they migrate it anyway, along with the duplicate tag archives and the one-off landing page from a conference that no longer exists. So before you touch a single template, score each page against four signals. Does it earn organic traffic or conversions? Does it hold inbound links or internal equity? Is the content still accurate? And is it just a worse copy of some other page you already have? Pages that score well on value and quality get migrated as-is. Value but poor quality: rewritten. Neither: killed and redirected. This happens in a spreadsheet, not a template (nobody wants to look at your spreadsheet, do it anyway). Skip it and you'll pay to haul dead weight into an expensive new platform, which is the one thing everyone swears they'll never do.
Why you score before you touch templates
Every page you migrate has a cost: it has to be mapped to a new content model, QA'd, redirected, and re-indexed. Teams that skip triage tend to lift-and-shift everything, which inflates the content model, multiplies redirect work, and buries genuinely important pages under years of expired promotions, duplicate tag archives, and one-off landing pages nobody remembers.
The scoring pass answers the only question that matters at the start: what actually deserves to exist on the new platform? Get that number down honestly and every downstream task, modeling, redirect mapping, QA, shrinks with it.
The four signals that drive the decision
Pull these for every URL. Most come from GA4 or your analytics tool, Google Search Console, a link tool (Ahrefs, Semrush, or GSC's links report), and a crawler.
- Value, traffic and conversions. Organic sessions, entrances, and any goal or revenue attribution over the last 12 months. Use a full year so you don't punish seasonal pages.
- Authority, inbound and internal links. External referring domains and internal links pointing in. A page with backlinks is a redirect obligation even if its traffic is low.
- Freshness and accuracy. Last meaningful update, plus whether the content still reflects current products, pricing, or policy. Old but correct is fine; old and wrong is not.
- Uniqueness. Is this the canonical answer for its topic, or a near-duplicate of a stronger page? Thin variants, faceted URLs, and legacy campaign clones cluster here.
The decision matrix
Cross value against quality (a blend of freshness, accuracy, and uniqueness) and the verdict falls out.
| High quality / unique | Low quality / duplicate | |
|---|---|---|
| High value (traffic, links, or revenue) | Migrate as-is | Rewrite / consolidate. Value is real, execution is not |
| Low value | Migrate only if it fills a topic gap; otherwise merge | Kill and redirect |
How to score each verdict in practice
Migrate
Reserve this for pages that earn their keep and don't need rework: money pages, evergreen guides with steady organic traffic, and anything with meaningful backlinks. These map cleanly to your new content model and should be your QA priority after go-live because breakage here costs the most.
Rewrite
Rewrite pages that attract demand but underdeliver: strong query impressions with weak clicks, outdated stats, or three thin posts that should be one authoritative page. A migration is the cheapest moment to consolidate, because you're already touching the URL and mapping the redirect. Fold the weaker URLs into the strongest one and point 301s accordingly.
Kill
Kill pages with no traffic, no links, no conversions, and no strategic role: expired events, retired product variants, duplicate tag and author archives, and parameter noise. Killing isn't just deleting, decide the redirect target for each. A relevant parent page beats a blanket redirect to the homepage, which Google frequently treats as a soft 404.
Edge cases the matrix doesn't cover
- Zero traffic, strong backlinks. Keep the URL alive (migrate or redirect) to preserve link equity, even if the content itself gets cut.
- High traffic, factually wrong. Rewrite before go-live, not after, migrating known-bad content re-publishes the error under fresh crawl attention.
- Seasonal spikes. A page that's dead 10 months a year may be critical for the other two. Look at annual peaks, not trailing 30 days.
- Pages that feed AI answers. If a URL is cited by AI engines or ranks for question queries, its structure and canonical status matter as much as its traffic. Killing or fragmenting it can quietly drop you out of generated answers.
How our scanner actually decides, since people ask
The matrix above is how a person should think about it. What the crawler does is narrower and more boring, and it is worth saying plainly because a scoring model that sounds sophisticated is easy to describe and hard to defend. There is no weighted percentage, no risk score out of a hundred. Four rules, and the first one that matches wins.
Kill comes first: the fetch failed, or the page answered with a 4xx or 5xx, or its content is genuinely duplicated somewhere else. One copy is kept and classified normally; the rest go.
Consolidate catches the same page reached twice — a trailing slash, a tracking parameter, a filter that produces another URL for content you already have, or a near-duplicate cluster member. The content survives. Only the surplus address goes.
Rewrite is for pages carrying survival flags: scripts, iframes, platform-specific markup that will not mean anything on the other side.
Migrate is what is left. Clean, reachable, unique.
Orphan status sits outside all four. A page nobody links to is a warning attached to whatever verdict it already earned, not a verdict of its own, because being unlinked tells you about the site's structure rather than about the page.
Turn the scores into a plan
Add one column to your inventory migrate, rewrite, or kill, and a second for the redirect target of every non-migrate URL. That single sheet becomes the spine of the project: the migrate list sizes your content model, the rewrite list becomes an editorial backlog, and the kill list becomes your redirect map. This is the automated triage that pre-migration content intelligence produces, crawling the whole site, joining traffic, link, and duplication data, and returning a scored plan you can argue with rather than a guess.
So here's what I actually think, plainly: do this before anyone models a single field, and do it while being ruthless still costs you nothing. Wait until templates exist and the deadline is breathing on your neck, and every page you never quite decided about quietly becomes a "migrate" by default. That's how the bloat wins. Score the pages now, and that expired fifteen-percent-off page gets a verdict instead of a free ride onto the new platform. You inherit a smaller, cleaner site. Not the same drawer of things nobody remembers, moved somewhere more expensive. Do your pre-migration triage first.