A redirect map is a sheet where every old URL sits next to the new URL that replaces it, built before you touch a server. Done well, it becomes the single source for your redirect rules, your tests and your post-launch monitoring. Done badly, it produces pile-ups on the home page, redirect chains and a wall of "not found" errors. Google’s site move with URL changes (Opens in a new tab) documentation treats URL mapping as a stage of its own and is blunt about the biggest mistake: do not redirect many old URLs to one irrelevant destination such as the home page of the new site.
If you prepare the URL list for a migration as an SEO specialist, hand a client's site to a new domain as an agency or rewrite your own shop's addresses, the sections that follow show how to build the sheet, which rule to apply to each row and how to turn the result into server rules.
Four outcomes for every old URL
One question decides every row: does the new site have a page that answers the same need as this old address? The answer leads to one of four outcomes.
| Situation | Decision | Expected response |
|---|---|---|
| A page on the new site covers the same topic | Map one to one | 301 or 308 |
| Several old pages were merged into one new page | Point all of them at the merged page | 301 or 308 |
| No page, but a category or hub covers the topic | Redirect to the closest hub | 301 or 308 |
| Nothing on the new site covers it | Do not redirect | 404 or 410 |
Google says merged pages should be redirected to the new combined page. If you merged two articles into one, the mechanics are in merging overlapping articles: 301 vs canonical. For content you are not moving, the documentation names 404 or 410, and the HTTP status codes guide covers the difference.
The tiebreaker rule
When a row feels uncertain, apply this test: if a visitor who arrives at the old address would land on a page that answers what they came for, redirect. If not, return a gone response. Practical suggestion: a redirect that exists only so the address "goes somewhere" misleads both the visitor and the search engine.
Worked rows
Example scenario: a recording studio rebuilds its site. The page "mixing and mastering" exists on the new site under a new path, which is a one-to-one row. Two old pages about "vocal booth" and "live room" now share one "studio rooms" page, which is a merge. A page for a workshop held three years ago has no counterpart and no value left, so it is a gone row. A page about the studio’s microphone collection is now a section of the "equipment" page, which makes it a hub row.
Where the old URL list comes from
Google names four sources for collecting old URLs. Each has a blind spot, which is why you combine them.
| Source | What it finds | Blind spot |
|---|---|---|
| Sitemaps and the Search Console Links report | Important pages and pages that receive links | Old pages missing from the sitemap |
| Server logs and analytics | High-traffic and recently visited URLs | Pages that attract links but few visits |
| The CMS | All content URLs | Parameter URLs and leftovers from earlier systems |
| Embedded resources | Images, videos, JavaScript and CSS files | Not visible in a page list |
Merge the four lists and remove duplicates. No single source gives a complete picture: a sitemap can lag behind the site, logs can miss old visits and a CMS forgets pages deleted years ago.
Where to start
Practical suggestion: begin with pages that receive external links in the Links report, then add the busiest pages from analytics. These two sets point at the pages that matter first and become the first rows you test. Complete the sheet with the CMS export afterwards.
Do not forget host variants
Google asks you to verify every variant of the old and new site in Search Console: www and non-www, http and https. The same logic applies to the inventory. Check which variants of the old site receive requests in the logs, because a list built only from canonical addresses does not show where requests for other variants will land. The www, https and trailing slash article explains how to choose one version.
Scrub the list before mapping
A raw export is not ready for mapping. Run these steps first.
- Collapse spellings of the same address: http and https, www and non-www, with and without a trailing slash.
- Group parameter URLs. Session IDs, tracking parameters and sort orders that do not change the content can be handled by one rule.
- Separate addresses that already redirect. Find the final target and connect it directly to the new destination, otherwise you create a chain.
- Separate addresses that return 404. If they have links or traffic, find a new equivalent.
- Flag addresses whose canonical tag points to a different page, and treat them together with the page they point to.
At the end you should hold a list of unique, real, rankable old addresses.
Why "everything to the home page" fails
The documentation's instruction is clear: do not redirect many old URLs to one irrelevant destination. Its example is the home page of the new site. The temptation arrives when a long list looks too tedious to map by hand.
Why it looks attractive
One rule covers thousands of rows and the launch checklist looks finished. The visitor, however, lands on a generic page rather than the product or article they wanted, and leaves. Google also gives you the better option for pages you are not moving: return 404 or 410.
Bulk rules that do work
Bulk redirects are not forbidden. Irrelevant targets are. When only the domain changes and paths stay the same, Google says a single wildcard server-side redirect may suffice. A block of old pages that all belong to one new category can sensibly go to that category.
| Bulk rule | Acceptable? |
|---|---|
| Domain change only, paths identical: one wildcard rule | Yes |
| All posts of an old category to the new category page | Only if that page covers the topic |
| All discontinued products to the home page | No, use 404 or 410 |
| Old products to the category of similar products | If the category answers the same need |
Keep an exceptions log
Occasionally a page without a counterpart still has links pointing to it. Look first for the nearest content page on the same subject. If none exists, record the 404 or 410 as a deliberate decision, with the reason, so nobody wonders later why it was not redirected.
Choosing the redirect mechanism
The Google page Redirects and Google Search (Opens in a new tab) separates permanent from temporary redirects. A permanent redirect, meaning 301 or 308, says a page has moved for good, and its target becomes the canonical URL. A temporary redirect, meaning 302, 303 or 307, keeps the source page as the one shown in results and does not signal that the target should be canonical. The page tells you to use permanent redirects when you are sure the redirect will not be reverted.
| Method | How Google treats it | Use for a move? |
|---|---|---|
| 301, 308 (server-side) | Permanent move, target becomes canonical | Yes, first choice |
| 302, 303, 307 (server-side) | Temporary, source page stays in results | No |
| Meta refresh, 0 seconds | Treated as permanent | Only if server-side is impossible |
| Meta refresh, delayed | Treated as temporary | No |
| JavaScript location redirect | Last resort. If rendering fails, Google might never see it | No |
The page shows examples for PHP, Apache and NGINX because implementation depends on the server. Server syntax varies by environment, so take the rule format from your host or administrator and connect it to the map. The page also says that after a domain move old URLs may keep appearing in results for a while, and that this fades over time without action.
The columns of the sheet
A spreadsheet is enough. A downloadable template file is outside the scope of this topic, but knowing the columns lets you build one in minutes.
| Column | Content | Why it is there |
|---|---|---|
| Old URL | Full address with protocol and host | Source of the rule |
| Source | Sitemap, log, analytics, CMS, links report | Shows how complete the inventory is |
| Priority | Traffic, links or business value | Decides what you test first |
| Mapping type | One to one, merge, hub, gone | Selects the rule |
| New URL | Full address, or empty for gone | Redirect target |
| Expected response | 301, 308, 404 or 410 | What the test should return |
| Test result | Real status, hops, final address | Verification record |
| Note | Reason, owner | Answers later questions |
Every row stands for one old address. If one old address appears twice, you have two conflicting rules and must resolve them before anything goes live.
Turning the sheet into server rules
There are three shapes a finished map can take. Which one you choose follows from the mapping types.
A single wildcard rule
If only the domain changes and every path stays the same, one rule moves the whole site. Testing is easy, but exceptions such as removed pages must be added to the rule separately.
Pattern rules
When paths change in a consistent way, for example every address under one old folder moves under a new folder, write one rule per pattern. Order the narrowest rule first and the broadest last. Because rule order changes outcomes, verify each new rule against your test rows.
A row-by-row list
Addresses that fit no pattern move from the sheet into a mapping list or a database table. Google notes that the pairs can live in a database or as rewrite rules. With very many rows, an application layer that reads a redirect list can be easier to maintain than a long server configuration, and that choice belongs with your system administrator.
No chains
Every target should be the final address. Google advises redirecting straight to the final destination and, where that cannot be done, keeping chains short: ideally three hops or fewer, and under five. For chains that already exist, read redirect chains and loops.
Testing the map
Google says to test redirects with URL Inspection for single URLs and with scripts or command-line tools for bulk testing. The "Test result" column is the record of that work.
- For each row, request the old URL and record the first response code and the redirect target.
- Follow the target and record the final response code, which should be 200.
- Count the hops. More than two suggests a chain.
- Compare expected and actual responses and mark every mismatch.
- Look for loops by finding rows whose target leads back to a form of the old address.
A quick chain check on the sheet itself
If a row’s target also appears as another row’s old URL, you have a chain. A spreadsheet formula that compares the target column to the old URL column finds these. A row whose target equals its own old URL is a loop. Catching both before the rules are written is far easier than debugging a live server.
Sampling
Thousands of rows cannot be checked by hand. Test all rows in the top priority group and a random sample from each mapping type. Run the same sample again after launch to show the rules still hold. A 200 on the target does not prove the content matches, so read the topic of a few target pages as well. For recrawling after the move, the tools in recrawling an updated page apply.
Updating the new site to match the map
The map is not only for redirects. It also drives the annotations on the new site, which Google lists as follows.
- Every new URL carries a self-referencing canonical tag, and canonicals point to new URLs.
- hreflang annotations use the new URLs.
- Internal links point to the new URLs.
- A new sitemap is saved and a list of sites linking to your old URLs is produced.
The "New URL" column feeds a bulk find-and-replace of internal links. Relying on the redirect while leaving internal links old turns every click into an extra hop, and Google notes that redirects are slower for users.
Extra work by move type
| Task | Domain change | Path change only |
|---|---|---|
| Canonical tag update | Yes | Yes |
| hreflang update | Yes | Yes |
| Internal link update | Yes | Yes |
| New sitemap | Yes | Yes |
| Change of address tool | Yes | No |
For the full migration picture see the website redesign SEO checklist, and for keeping link value through the move read preserving backlink value during site migration.
Example scenario: a recording studio
Example scenario: a recording studio with two rooms and a small shop for sample packs moves to a new domain and trims its page structure. The old site grew over years, with a booking page, equipment pages, blog posts about sessions and a calendar of past events.
The inventory comes from the sitemap, analytics, the Links report and a CMS export. After removing duplicates and host variants, the equipment pages need extra care, because several were merged. The two room pages become one "rooms" page, so both old addresses point to it. Past event pages have no counterpart; the studio returns 410 for them and keeps a short archive section on the blog instead. The booking page is mapped one to one and tested end to end, since a redirect that breaks the form would cost real bookings.
| Old address type | Mapping type | Response |
|---|---|---|
| Booking page | One to one | 301, form tested separately |
| Two room pages | Merge | 301 to the combined page |
| Microphone collection page | Hub | 301 to the equipment page |
| Sample pack pages | One to one | 301 |
| Past events | Gone | 410 |
A week after the move the studio reruns the sample and finds one product image that nobody had listed, because the image appeared in neither the sitemap nor analytics. It joins the sheet with its new location, and the rule is updated.
Row patterns by URL type
On a real site the sheet consists of a few recurring URL types. Recognizing the type speeds up mapping. The table gives a first decision for each type. The decisions are practical suggestions.
| URL type | First decision | Question to ask |
|---|---|---|
| Home page | 301 to the new home page | Is there a language or regional version? |
| Category page | 301 to the new category | Was the category merged or split? |
| Product or article page | One to one 301 | Is the content still published? |
| Paginated list | The same page of the new list, or its first page | Does pagination work the same on the new site? |
| Tag and archive pages | The new tag page, or removal | Does the new site have this page type? |
| Images and PDF files | 301 to the new file location | Did the file name change? |
| Print or other derived versions | The main page equivalent | Is the variant still produced? |
| Test and temporary pages | Remove, 404 or 410 | Does anything link to it? |
Google’s pages used here set no separate rule for pagination. If your new site has the same list page, map to it. Otherwise the first page of the list is the nearest content for the visitor.
More than one candidate
An old page may have two candidate targets. Choose by visitor intent: if the old page sold something, send it to the page that sells it, and if it solved a problem, send it to the page that explains the solution. Write the reason in the note column so a new teammate does not reopen the question months later.
Running a large map
As rows multiply, the sheet starts to behave like a software project. Three habits protect it.
One master copy
Keep a single master with dated changes. If two people edit separate copies, one old address can end up with two different targets. Assign an owner to each row or group of rows.
Cut it into sections
Google advises moving large sites one section at a time. The sheet can follow the same logic: finish and test one section, then proceed to the next. Draw the section boundaries in the rules as well, so a rule for one section does not capture addresses from another.
An approval gate
Before going live, let the business owner review the rows at the top of the priority column. Finding a wrong target for a busy page now costs much less than finding it after the rule is live. One extra column asking whether the target page answers the topic, with a yes or no, is enough.
Keeping the sheet alive after launch
Google points to Search Console reports, server logs and analytics for monitoring. The sheet is the reference for all three. When an old address shows as not found, look in the sheet first: if the row exists, ask why the rule fails, and if not, add the row. Add addresses that Googlebot requests but the sheet lacks. Before retiring a redirect, check whether the old address still receives traffic. Google recommends keeping redirects as long as possible, with at least one year as the benchmark, and time limits in its documentation can change, so confirm the current page.
Checklist before handing the sheet over
- The inventory came from the four sources, was de-duplicated and host variants were cleaned.
- Every row has a mapping type, a new URL and an expected response.
- No row points at the home page, or the exception is written down with a reason.
- Pages that are gone have a recorded 404 or 410 decision.
- Targets are final addresses and there are no chains.
- The redirect type is permanent, 301 or 308.
- Canonicals, hreflang and internal links were updated on the new site.
- Test results are recorded in the sheet.
Mistakes to avoid
Relying on a single source
A list from the sitemap alone or the CMS alone misses linked pages that neither knows about.
Not checking the target
A 301 is not enough. Confirm that the target returns 200 and covers the right topic. Google lists redirects to non-existent URLs as a problem seen in moves.
Using a temporary redirect
For a permanent move use 301 or 308 rather than 302. A temporary redirect does not make the target canonical.
Leaving internal links untouched
If the redirect works, the old internal links seem harmless, but each costs visitors an extra hop.
Closing the sheet on launch day
The sheet is a living document. Add newly found old addresses after the move and update the rule for each new row.
Frequently Asked Questions
What is a redirect map?
It is a list that pairs each old URL with its new equivalent. Google’s move documentation treats collecting old URLs and storing the pairs, in a database or as rewrite rules, as a step of the move plan.
Where do I collect the old URLs?
Google names sitemaps and the Search Console Links report, server logs and analytics, the CMS, and embedded resources such as images, videos, JavaScript and CSS. Combine them, since no single source is complete.
Can I redirect pages without a counterpart to the home page?
No. Google tells you not to redirect many old URLs to one irrelevant destination such as the home page. For content you are not moving, return 404 or 410.
Should I use 301 or 302?
For a permanent move use 301 or 308. Google explains that a permanent redirect makes the target canonical, while a temporary one keeps the source page in results.
How long may a redirect chain be?
Google advises redirecting straight to the final destination. If that is impossible, keep chains short: ideally three hops or fewer, and under five.
How do I test the map?
Google suggests URL Inspection for single URLs and scripts or command-line tools for bulk testing. Record each row’s status code, hop count and final address in the sheet.
If only the domain changes, must I map every address one by one?
Google says a single wildcard server-side redirect may suffice when only the domain changes. You still need to list removed pages separately.
Conclusion
The redirect map is the quiet backbone of a move: rules, tests, internal link updates and monitoring all feed from it. Collect old addresses from several sources, remove duplicates and tie each row to one of four outcomes: one to one, merge, hub or gone. Avoid bulk redirects to irrelevant targets, use permanent redirects, point targets at final addresses and record your test results in the sheet.
When the new addresses are live and you want their first mentions, our guest post service may help.
Post a comment
Please log in to leave a comment. Login