A redirect map can return a perfect 301 for every old URL and still damage the website. The status code moves the request. It cannot decide whether the destination preserves the old page’s purpose, evidence or next action.
That is why I treat migration mapping as content architecture, not plumbing. The question is not merely “where should this URL go?” It is: where can a person continue substantially the same task? If the spreadsheet cannot answer that, a green status column is giving false reassurance.
TL;DR
- A technically valid redirect can still be wrong. URL preservation is not job preservation.
- Decide whether the page has actually moved. An honest 404 or 410 can beat a homepage redirect that claims continuity.
- Map the old job before choosing the new URL, and grade each replacement Full, Partial or None.
- Test two columns, protocol correctness and task continuity. A row fails while either fails.
A technically valid redirect can still be wrong
Google describes permanent redirects as signals that a destination should become the canonical URL shown in Search. Google also states that permanent redirects do not cause a loss of PageRank during a site move (Google: redirects and Search; Google: site moves). That does not make every destination equally useful.
An old boiler-repair page redirected to the heating homepage has moved successfully at the protocol level. The visitor may still have lost the repair information, service conditions and route to request help. The server completed the instruction. The migration failed the task.
This is where projects go wrong. Old URLs arrive late. The spreadsheet is treated as a developer deliverable. Awkward pages are mapped to the strongest-looking category or homepage because somebody has to fill the destination cell before launch. I call this URL preservation without job preservation. It creates the appearance of a controlled move while deleting the reason the old route mattered.
Decide whether the page has actually moved
A permanent redirect is appropriate when the old destination has genuinely been replaced: the same page has a new address, overlapping resources have been consolidated, a product has a true successor or a domain has moved. It is not appropriate merely because the old page is inconvenient. Use four outcomes:
| Old-page condition | Defensible action | What must be true |
|---|---|---|
| Same job, new address | Permanent redirect | The destination performs substantially the same task |
| Several pages consolidated | Permanent redirects | The new page carries the useful work of each admitted source |
| Temporary interruption | Temporary response or redirect | The original URL remains the intended long-term address |
| No honest replacement | 404 or 410 | No destination can continue the task without misleading the visitor |
The final row makes teams uncomfortable because a missing-page response looks like failure. Sometimes it is the honest result. A homepage redirect can be worse because it claims continuity that does not exist.
Do not hide a content decision inside a status code.
Map the old job before choosing the new URL
A URL string rarely explains what the page did, especially after years of CMS changes, acquisitions and inconsistent naming. For every material route, record:
I use three replacement verdicts. Full: the destination continues the essential task and retains the material conditions. Partial: the destination is related but loses information, evidence or action the old page supplied. None: the destination does not meaningfully continue the task.
Partial is a warning, not an automatic approval. It may justify improving the destination, keeping the source, merging only part of its content or accepting removal. It does not mean relevance can be manufactured by making the redirect permanent.
The strongest page is not necessarily the closest replacement
Migration teams often confuse authority with continuity. The homepage may have the most links. A broad service category may rank best. Neither automatically replaces a specialist page. Compare source and destination across five dimensions:
| Dimension | Continuity question |
|---|---|
| Audience | Is the destination for substantially the same person and situation? |
| Intent | Does it answer the same question or enable the same transaction? |
| Scope | Are the service, product, geography and eligibility conditions preserved? |
| Evidence | Does the destination retain the proof or documentation needed for the decision? |
| Action | Can the visitor still complete the appropriate next step? |
I call this the Task Continuity Matrix. A mismatch does not always block a redirect, but it must be resolved consciously. A discontinued product may redirect to a true successor when the relationship and differences are explained. It should not automatically land on a category containing hundreds of unrelated items. A branch page may move to a regional service page if that page genuinely carries the territory, service and contact route the visitor needs. “Related topic” is too weak. The replacement must complete enough of the old decision to deserve the request.
Preserve the estate before changing it
Do not build the migration map from current navigation alone. Navigation shows the routes the current template remembers, not the complete public estate. Collect:
- the old XML sitemap;
- a crawl of the live site;
- Search Console page data;
- analytics where appropriate;
- known external links;
- campaign destinations;
- downloadable files;
- legacy paths still used by customers, staff or documentation;
- URLs in canonicals, hreflang, structured references and internal links.
This evidence changes decisions. A quiet page with valuable external references may deserve consolidation. An apparently obsolete file may still support customers. A route absent from the developer’s list may expose a migration gap before users find it. Preserve a dated, recoverable copy of the approved map. A plugin screen is not a durable migration record. If performance changes later, the team needs to know which approved decision produced each destination.
My preference is to record the evidence without letting raw traffic determine value. A low-traffic page may still be operationally essential. A high-traffic page may deserve removal if it attracts the wrong task and has no honest successor.
Consolidation must carry the useful work forward
Several old pages can legitimately become one new page, but “we merged them” is not proof that the destination inherited their value. For each source, identify the distinct question answered, the useful evidence carried, the material limitation, any unique example or tool, the next action, and the reason the destination can absorb it without becoming incoherent.
Then inspect the proposed consolidated page. If three sources each carried a different eligibility condition and the destination retains only the broad sales copy, the migration has not consolidated their work. It has deleted it behind three redirects.
The Content Inheritance Test: every admitted source must contribute something visible to the destination or have an explicit reason its old material is no longer valid or useful.
This prevents “consolidation” becoming a polite name for a content purge.
Migration spreadsheet full of green 301s and homepage destinations? We can grade every row for task continuity before it goes live.
Remove detours from the new website
Redirects protect old requests. They should not become the new site’s navigation system. After implementation, update navigation, contextual links, canonicals, hreflang references and sitemaps to point directly to the final URLs. A visitor clicking a new-site link should not travel through a historical address first. Google recommends sending users directly to final destinations and avoiding redirect chains during site moves (Google: site-move guidance). Chains add latency, complicate testing and create more places for old decisions to collide with new ones. Flatten known chains where it is safe. Keep the original legacy route functioning, but point it directly at the final maintained page.
Test the response, page and meaning
A migration test that checks only for 301 is incomplete. My acceptance test has two columns. A row does not pass while either one fails. For every priority source, verify:
Protocol correctness
Task continuity
Wildcard rules deserve extra suspicion. They can solve thousands of predictable paths efficiently. One loose assumption can also misroute thousands of pages with impressive consistency.
Judge the migration by destination stability
The first post-launch report should not celebrate the number of rules installed. It should show whether important old routes reach relevant pages and whether those pages remain available, indexable and useful. Monitor unexplained missing-page responses; old URLs still referenced internally; redirect chains introduced by later changes; destinations that stop returning healthy content; unintended canonical or indexation changes; sections losing visibility out of proportion to the rest of the site; and priority tasks that no longer convert or complete.
Temporary movement can occur while search systems recrawl and process a move. Document material changes before reacting to every daily fluctuation. Investigate persistent, section-specific discrepancies against the approved map. The stop condition is practical: priority old URLs resolve directly to approved replacements, new references use final URLs, important destinations remain healthy and no unexplained task loss is still tied to the migration.
Make every destination defend itself
Before approving a row, require the person proposing it to finish this sentence:
This old URL moves here because the destination now performs the same job by…
If nobody can finish it, the mapping remains unresolved. Improve the destination, retain the source, choose another route or return an honest missing-page response. A redirect can preserve an address change. It cannot transfer missing evidence, replace a deleted decision or turn a broad page into a useful successor.
Map the job. Test continuity. Preserve the evidence. Then let the server perform the move the organisation has actually decided to make.
Good to know
FAQs
Rarely. A homepage redirect moves the request but usually abandons the visitor’s task, and it claims a continuity that does not exist. If no destination can continue the job, an honest 404 or 410 is often the better result.
Google states that permanent redirects do not cause a loss of PageRank during a site move. That is a protocol answer. It says nothing about whether the destination preserves the old page’s purpose, evidence or next action.
A five-dimension comparison of source and destination: audience, intent, scope, evidence and action. A mismatch does not always block a redirect, but it must be resolved consciously rather than hidden behind a status code.
Two columns. Protocol correctness: response, Location header, healthy final page, no avoidable chain, indexable and canonically coherent. Task continuity: the old task is still possible, scope is visible, actions work and mobile visitors get a usable journey. A row fails while either column fails.
Next step
Is your redirect map preserving jobs or just URLs?
We map the old page’s job, grade every destination Full, Partial or None, and test protocol correctness and task continuity side by side before launch.