When we replatform a website, the code is the easy part. The URLs are the dangerous part. A business that has been online for years has equity spread across every address it ever published: backlinks from suppliers and press, old directory listings, bookmarks, and thousands of entries in search indexes. Break those addresses and rankings that took years to earn start bleeding away, quietly, with no error message.
So we treat legacy URLs the way an accountant treats money in transit. Every one is inventoried, every route is mapped, and nothing is allowed to fall between systems. A 301 redirect is a statement that the page moved permanently, and search engines transfer the signals they had accumulated to the new address. A 404 is a dead end, and nothing transfers through a dead end.
One hop, exactly
Kwaan Bear Technology is our cleanest example. The client is an Alaska Native Corporation federal IT contractor whose previous site ran on WordPress. When we rebuilt kwaan.tech, the redirect layer shipped as first-class engineering: 23 exact legacy-to-new mappings, a subtree rule, and a generic trailing-slash collapse, all issuing real HTTP 301s from middleware.
The detail worth stealing is the hop count. Frameworks love to chain redirects. An old trailing-slash WordPress URL hits the framework's automatic 308, which then lands on our 301, and now a crawler crosses two hops to reach the content. We turned the framework's trailing-slash handling off and folded that logic into the same middleware, so every legacy URL resolves in exactly one hop. The config file documents why the setting exists, so nobody flips it back.
We redirected the old sitemaps too. Five legacy WordPress sitemap URLs, the kind Yoast generates, were still in the search engines' memory, so they 301 to the new sitemap instead of 404ing. Query strings are preserved through every redirect. And after cutover we verified the whole map against the live site: thirty URLs, each answering with a single 301 to the right destination.
Why obsess over one hop? Because every extra hop adds latency for real visitors, spends the crawl budget search engines meter out per site, and adds one more place for a chain to break in a future refactor. A one-hop rule is also testable: you can script a check that requests every legacy URL and fails if any response chains. Rules you can test are rules that survive.
Redirect maps are data, not config
On AAS and Creative Maintenance Solutions, the redirect map lives as a typed data file that the framework config imports. Legacy sections like /what-we-do and /who-we-are carry wildcard rules, so every child URL inherits its mapping without being listed one by one.
Keeping the map as data pays off in review. You can read it top to bottom, diff it in version control, and reason about it like any other content. It also holds process. The Creative Maintenance map carries a written pre-cutover reminder to pull the indexed-URL list from Search Console, because the URLs a search engine actually remembers are the ones that matter, not the ones you assume it does. And the map stays alive after launch: when the client remembers an old campaign URL months later, it becomes one more typed entry and a one-line diff, reviewed and deployed like any other change.
The best redirect is no redirect
On Prime Coat we sidestepped the problem entirely. The rebuild kept slug parity with the existing site: every page kept its exact URL, so no redirect layer was needed at all. That included all 64 blog posts, carried over verbatim at their original addresses. We went further and preserved each page's meta title and description from a crawl of the source site. To a search engine, the replatform is nearly invisible, which is precisely the goal.
Parity is not always on the table. Information architecture improves, sections merge, and old CMS URL patterns deserve retirement. On a build for a healthcare enterprise, 19 permanent redirects map the superseded structure onto the new one. On Belzona Baton Rouge, addresses from a legacy .aspx platform still 301 home. But when you can keep the URLs, keep the URLs. It is the cheapest insurance in SEO.
The checklist we run
- Inventory every URL the old site ever answered: crawl it, pull its sitemap, and check what search engines actually have indexed.
- Map each one to its new home with a real 301, not a 302 and not a meta refresh.
- Collapse chains so every path resolves in one hop, including trailing-slash variants.
- Redirect the old sitemap URLs themselves; crawlers keep requesting them for months.
- Preserve query strings through every redirect.
- Verify the whole map against the live site after cutover, not just in development.
Two failure modes account for most of the equity lost in replatforms. The first is the unmapped URL: a page the new team never knew existed, findable only by crawling the old site and pulling what the index actually holds. The second is the lazy chain: redirects that technically arrive but cross two or three hops, wasting crawl budget and weakening the signal on every pass.
None of this work shows up in a design review, and nobody screenshots a redirect. But a replatform that keeps its rankings instead of restarting them is the difference between a new site that accelerates a business and one that quietly costs it a year. Years of link equity is an asset. Protect it like one.
