Site Rebuilds and Domain Changes: How to Draw a 301 Handover Map
1. The most expensive mistake in a rebuild is a missed URL in the handover
Most rebuilds fail not because the new site looks bad, but because the moment the old site was replaced, a batch of URLs quietly broke. Broken pages don’t throw an error, they just stop getting traffic, and it never comes back.
First, be clear about which kind of change you’re actually making. The three types carry very different levels of risk:
| Rebuild type | What changes | Traffic impact range | Recovery time | Reversible |
|---|---|---|---|---|
| Redesign only, URLs unchanged | Templates, styles, section placement | Down 5% to 15% | 2 to 4 weeks | Can roll back |
| Change URL structure | New directories, new slugs, new pagination rules | Down 15% to 40% | 6 to 12 weeks | Needs 301 fallback |
| Change domain | Primary domain changes | Down 20% to 45% | 8 to 16 weeks | Only 301 can save you |
| Change domain and structure | Both of the above at once | Down 30% to 60% | 12 to 24 weeks | Extremely hard to roll back |
The ranges in this table come from observations on typical projects, not promises. The point it makes is one judgment call:don’t change the domain and the structure at the same time. If the business requires doing both together, accept a fluctuation period of 12 weeks or more, and prepare an expectation brief your boss can actually understand.
Among missed handover items, four are by far the most common, and they cost you differently:
| Missed item | Consequence | Recovery time | Can it be fixed |
|---|---|---|---|
| Detail pages not 301’d | Pages vanish completely, accumulated authority resets to zero | Original rankings can’t be recovered | Only rewrite and resubmit |
| 301 points to the homepage | Treated as a soft 404, link authority is discarded | 4 to 8 weeks | Fix the target and it recovers gradually |
| Redirect chains longer than 3 hops | Crawl budget wasted, authority decays with each hop | 2 to 6 weeks | Can be fixed directly |
| Filter parameter pages not handled | Large numbers of near-duplicate URLs eat up crawl quota | 4 to 12 weeks | Can be fixed |
The second row deserves the most caution. The idea that “they’re all my pages anyway, a 301 to the homepage beats a 404” is wrong in practice: search engines treat a large batch of redirects pointing to the homepage as soft 404s, link authority doesn’t transfer, and the new homepage’s anchor text signals get muddled.
2. Step one: build a URL balance sheet
Before you touch anything, export every URL from the old site. “Every” means the union of three sources: server access logs, the sitemap, and the list of URLs already indexed by search engines. Exporting the sitemap alone isn’t enough — there are always pages that aren’t in the sitemap but still have external links.
Keep at least these eight fields when you export. Miss one and you’ll be redoing the work later:
| Field | Source | Purpose |
|---|---|---|
| Full URL | Logs and sitemap | Primary key of the mapping table |
| Monthly organic traffic | Analytics tool | Priority decision |
| Number of referring domains | Backlink tool | Priority decision |
| Target leads or conversions | Forms and analytics | Priority decision |
| Page title | Reference for the matching content on the new site | Match to new page |
| HTTP status code | Bulk check | Remove dead URLs |
| Language and region tag | hreflang or directory structure | Required for multilingual sites |
| Action | Manual decision | Keep, map, merge, retire |
The last column is the heart of this table. Tag every URL with one of four labels, and write the criteria in stone — no gut calls:
| Action | Test | Technical action |
|---|---|---|
| Keep | The new site has a page at the same path with essentially the same content | Leave the URL as is, no redirect |
| Map | The new site has matching content but the path changed | One-to-one 301 |
| Merge | Several old pages overlap in content and the new site keeps only one | Many-to-one 301, keeping the page with the most complete content |
| Retire | The content is genuinely gone, with no backlinks and no traffic | Return 410, no 301 |
There is only one order of priority. Don’t flip it:Backlinks first, then traffic, then conversions. A page with zero organic search traffic but three external links may be worth more than a page with 20 visits a day and no backlinks, because backlinks are a transferable asset while traffic fluctuates with algorithms and seasons. Any page that fails all three tests — no backlinks, no traffic, no conversions — gets marked for retirement. These pages usually make up 30% to 50% of the old site’s URLs, and clearing them out is itself a win from the rebuild.
When this approach doesn’t apply: If the old site has been live for less than 12 months, or its cumulative organic traffic has stayed below 500 visits a month, then almost every value in the “balance sheet” is zero and there’s no basis for setting mapping priorities. In that case you don’t need the full analysis — just make sure of three things: every URL that was ever promoted, advertised, or printed on materials must 301; every URL in the sitemap must 301 or 410; the rest can simply settle into the new site’s structure.
3. Three types of mapping rules: one-to-one, many-to-one, one-to-many
There are only three types of mapping rules, and each has its own clear method and its own pitfalls.
| Mapping type | Trigger | How it’s written | Main risk |
|---|---|---|---|
| One-to-one, same path | Path changed but the content maps to exactly one page | 301 the old URL to the new URL | Mismatched slug spellings between old and new cause missed mappings |
| Many-to-one merge | Three or more old pages overlap in content | 301 them all to the page you keep | Picking the wrong page to keep and losing the most valuable one |
| One-to-many split | One 3,000-word article split into three pages | 301 old URLs to the most relevant new page | Two other pages lose their entry points and need internal links |
Many-to-one consolidation is the easiest type to get wrong. Consolidation isn’t about dumping traffic onto any page; the decision criteria have an order:
| Rationale | Weight | Description |
|---|---|---|
| Most external links | High | Keep whichever page external links point to |
| Most complete content | Medium | The page with the most complete specs, case studies, and FAQ |
| Best existing rankings | Medium | Keep the page with stable rankings |
| Shortest URL | Low | Use when the first three are tied |
One-to-many splits require attention to the reverse entry-point problem. Of the three split pages, only the one that gets the 301 inherits the old links; the other two start from zero. The fix is to add an intro paragraph at the top of the kept page with internal links to the other two, and monitor their indexing status for two months.
4. Four high-risk areas to check one by one
Four areas repeatedly cause problems during restructuring. Their common trait: no one notices when things go well, but troubleshooting costs are extremely high when they go wrong.
| High-risk area | Typical incident | How to check | Pass criteria |
|---|---|---|---|
| Multilingual URLs and hreflang | Language versions point to each other incorrectly or all point to the default language | Crawl the page source of each language version | Each language’s hreflang points to the corresponding page on the new domain |
| Pagination and filter parameters | Parameter pages get heavily indexed, or the pagination chain breaks | Test 20 URLs with parameters | Parameter pages return the canonical page or 410, and the pagination chain works end to end |
| Attachments and PDFs | Quotes, drawings, and catalogs return 404 | Export all attachment URLs and probe them in bulk | All valid attachments 301 to the new addresses |
| Forms and short links | Campaign short links and trade show QR codes stop working | Test every URL used off-site one by one | All reachable and forms submittable |
Multilingual sites need a separate checklist because they have the highest error rate:
| Checklist | What not to do | The right way |
|---|---|---|
| hreflang targets | All point to the old default language domain | Each language points to the corresponding language page on the new domain |
| Language directories | New site switches to parameter-based selection, e.g. adding a query parameter | Keep separate directories or subdomains, one URL per language |
| Sitemap | Only submit the default language sitemap | Submit per language and mark language alternates in the sitemap |
| Indexed pages | Only 301 the default language | 301 each language’s old URLs separately |
The attachment area is also often overlooked. Files like catalogs, quotes, and drawings are often targets of external links and are the most forwarded items by overseas buyers. Bulk-probing status codes takes less than an hour, but missing them means permanently losing a batch of external links.
5. Execution order and backup points before irreversible operations
Wrong order doubles the workload. The following seven steps are a fixed sequence; don’t adjust them just because you’re in a hurry.
| Steps | Action | Deliverables | Reversible? |
|---|---|---|---|
| Step one | Back up all old site data and files | Database and file archives, saved to a local copy as well | Reversible, this is the fallback |
| Step two | Export and tag the URL inventory | A table with the action for each URL | Reversible |
| Step three | Build the full structure on the new site and add the content | The new site is reachable on a temporary domain | Reversible |
| Step 4 | Test every 301 one by one on the temporary address | Mapping table test log | Reversible |
| Step 5 | Switch DNS and put the old site into read-only mode | Screenshot of the DNS records | Irreversible — Step 4 must be confirmed complete |
| Step 6 | Submit the new sitemap and keep the old sitemap live for 60 days | Both sitemaps are reachable | Reversible |
| Step 7 | Turn on the 301 rules and watch the logs | Daily redirect hits and error counts | Reversible |
Step 5 is the only irreversible operation. Before you switch DNS, confirm three things: the backup is done and can be restored, every entry in the mapping table has passed testing on the new site’s temporary address, and the read-only copy of the old site can still be switched back at any time. If any one of these is not done, don’t switch.
There are six common mistakes in how redirects are written. Most are configuration issues, but the impact is significant:
| Wrong approach | Consequence | The right way |
|---|---|---|
| Chained redirects: A to B to C | Authority decays at every hop and crawling slows down | Point everything straight to the final URL |
| 301 everything to the homepage | Treated as a soft 404, so no authority is passed | Map each URL one-to-one to its counterpart page |
| 301 to a URL that already returns 404 | The redirect chain breaks, which is the same as a broken link | Verify the target URL returns 200 first |
| Mixing 302 and 301 | 302 does not pass authority | Use 301 for every permanent migration |
| Rules only cover the version with a trailing slash or the one without | The other half of the URLs break | Configure rules for both forms |
| Using an in-page redirect instead of a server redirect | Crawlers can’t see the real redirect | Configure it at the server or application layer |
6. What to watch in the 30 days after the switch, and when not to redirect
After the switch, don’t check rankings every day. Four metrics are enough, and one of them is your stop-loss line.
| Metrics to watch | Normal range | Stop-loss line | What to check first when you hit the stop-loss line |
|---|---|---|---|
| Pages indexed on the new site | 70% or more of the old site’s count within 30 days | Below 40% | Sitemap, robots, redirect chain |
| Crawl frequency | Rises 20% to 50% in the first two weeks, then falls back | Still below the old site’s level after two weeks | Server logs, response time, status codes |
| Rankings for core keywords | Fluctuating 3 to 10 positions is normal | Drops out of the top 50 and doesn’t recover within 4 weeks | Page content consistency, internal links |
| Inquiries per week | Fluctuating 10% to 30% is normal | Below half the old site’s level for 3 weeks straight | Form functionality, redirect hit logs |
Note the third row: rankings bouncing between positions 3 and 10 is normal after a rebuild, and you don’t need to intervene for two months. What does need action is a batch of pages disappearing at the same time — that usually means a missing mapping, not an algorithm change.
Finally, the reverse case: three types of pages should not get a 301 and should return 410 instead.
| Situation | Why not 301 | Correct status code |
|---|---|---|
| The content really is permanently gone and there’s no replacement | A 301 to the homepage is judged a soft 404 | 410 |
| Taken down, with no backlinks or traffic | Keeping the redirect brings no benefit and only adds maintenance | 410 |
| Large numbers of low-quality parameter pages | Redirecting keeps duplicate URLs indexed | 410, plus a canonical tag |
The difference between 410 and 301 isn’t a technicality, it’s semantics: 301 means the content moved elsewhere, 410 means the content no longer exists. 301-redirecting a large batch of dead pages to your homepage tells search engines “all this content moved to the homepage,” which doesn’t hold up on judgment, and the cost is that your homepage’s link signals get diluted. For the related delivery process and checklist, see /site/ for the delivery notes, or just follow /contact/ to send us your URL list for a mapping check.
FAQ
Final thoughts
In a rebuild, the time spent on design is often three times the time spent on the handover, but the latter determines success or failure. Design can be changed next year; once a URL breaks, the accumulated authority for that page is gone permanently, with no second chance to recover it.
Turn the handover into a table, a self-check list, and a pre-switch test of every item, and this no longer depends on luck. The only place that truly requires judgment is this: for each URL, should it be kept, mapped, merged, or discarded? Once that judgment is made, the rest is execution.
One-line takeaway: a rebuild isn’t scary because it looks ugly; it’s scary because after the rebuild, no one can clearly say where every old URL went.
