The short answer
A redirect map is one row per old URL, pointing straight at the new URL that replaces it, turned into permanent server-side redirects on launch day. Use a 301 or 308. Match by meaning, so a pricing post goes to a pricing page, not the home page. Send every old URL straight to its final address, with no chains.
Google's site move guide is plain on all of this, and there is no clever shortcut. The work is in the care you take row by row. Our free redirect map tool does the first matching for you in your browser, so you can spend your time on the rows that need a person.
Decide what happens to each URL
Every row from your inventory gets one of four outcomes.
| Outcome | When | What the server does |
|---|---|---|
| Same URL | The page stays at the same address | Nothing. No rule is needed. This is the best case. |
| Moved | Same content at a new address | 301 to the new URL |
| Merged | Several old pages now live in one new page | 301 each old URL to that one page |
| Retired | You are not moving it, and nothing else covers the topic | Return 404 or 410 |
The merged and retired rows are where people go wrong. Google says not to redirect many old URLs to one irrelevant page, such as the new home page, and that URLs you are not moving should return a 404 or 410. Consolidated content is the exception: if old pages really were combined into a new one, redirect them to it.
Decision rule Would someone who clicked the old link be happy to land here? If yes, redirect. If no, let it return a 404 or 410, with a helpful 404 page. Do not send it to the home page to feel safe.
Which status code
The code tells search engines whether the change is for good. From Google's redirects documentation:
| Type | How Google treats it | Use it for |
|---|---|---|
| 301, 308 (server) | Permanent. The target is treated as the canonical URL. | Every row in your map |
| 302, 303, 307 (server) | Temporary. It follows them, but does not take them as a sign the target should be canonical. | Short-term changes. Not a site move. |
| Instant meta refresh | Treated as permanent | Only when your host offers nothing better |
| Delayed meta refresh | Treated as temporary | Avoid |
| JavaScript redirect | Can be read as permanent, but only if rendering works | Last resort |
Check what your platform really sends. Some tools and plugins say "redirect" and send a 302, so test the status code, not the setting.
Clean up the addresses
Two URLs that look the same to you may be different to a server. Decide these once, and write them at the top of the sheet.
- Trailing slash. Pick one form.
/about/and/aboutare different URLs. - Capitals.
/About-Usand/about-usmay be different on your old host. - Host and protocol. Pick one of
wwwor non-www, andhttps. Send the others straight to it. - Query strings. Tracking codes need no rules of their own. Filtered or paged URLs need a decision.
- Files. PDFs and images with links from other sites keep their address, or get a row.
- Languages. Map each language on its own, and update hreflang to the new URLs.
No chains and no loops
If an old URL redirects to B, and B now redirects to C, send the old URL straight to C. Your old server may already hold years of redirects, so go through them too, and update each to the final address.
Google's site move guide says Googlebot can follow up to 10 hops, but advises redirecting straight to the final destination. If you cannot, it asks you to keep the chain short, ideally no more than 3 and fewer than 5. A chain also slows people down. Loops, where two rules point at each other, simply break the page.
| Before | After |
|---|---|
/old-pricing to /pricing/ to /pricing | /old-pricing to /pricing |
http://site.com/a to https://site.com/a to https://www.site.com/a to https://www.site.com/b | http://site.com/a to https://www.site.com/b, where your host allows it |
/a to /b, and /b to /a | Pick the real page, and remove the other rule |
A shortcut: the redirect map tool
The redirect map tool takes your old list and your new list as plain URLs, CSV or sitemap XML. It runs only in your browser and sends us nothing.
- Paste the old URLs and the new URLs.
- It matches them and marks each row exact, high, medium, low or none.
- It flags chains, loops, duplicates and rows where nothing changed.
- Read every medium, low and unmatched row, and change what is wrong.
- Export a Webflow CSV, a
vercel.jsonblock, a_redirectsfile, or a plain CSV.
A matching tool gets you a draft. It reads addresses, not meaning, so a person still has to read the rows that matter. Treat every tier A page as needing your own eyes, whatever the confidence says.
Rules, patterns and limits
Use one exact rule per URL for your tier A and B pages. Patterns are fine when a whole folder moves the same way, such as /news/* to /blog/*, but check the exceptions first.
Check your new host's limits before you plan on 5,000 single rules. Webflow, Vercel, Netlify and Cloudflare each handle redirects in their own way, with their own limits and formats, and those limits vary by plan and change over time. Read the current docs for your plan. If a list is too long, group by pattern, or handle it at a layer in front of the site.
Our redirect map template has the sheet columns if you prefer to start by hand.
Test before launch
Do not wait for launch day to find out the map is wrong. Most of this can be done before the switch.
- Every new URL in the map exists and returns 200 on staging.
- No new URL in the map is itself an old URL with another rule.
- No two rows share the same old URL.
- Rules load on staging if your host allows it, and a script or crawler checks a sample.
- Tier A rows are checked by a person: is this the right page?
- One person signs off, and a dated copy of the map is saved.
Done when Launch only when every old URL with clicks or links has an answer that a person has checked, and no row is still blank.
On launch day you run the same check against the live site, using your whole old URL list. Chapter 5 covers that.


