Major matching improvements for multilingual and multi-market sites: locale-aware routing, more accurate confidence, and better product matching.
Improved
- Localized URLs now map to their own market. When a source URL is for a specific language or region, it now resolves to the matching localized page instead of falling back to the global page, whenever a localized version exists.
- More accurate confidence labels. A match where the path is the same but the market or region changed is now shown as high confidence rather than a 100% exact match. The exact tier is reserved for genuinely identical pages.
- Better product matching. The engine now considers a wider set of candidates, so it finds the right product even when the slug was renamed or translated across languages.
- Uncertain matches are flagged for review. When a product is no longer on the new site, the closest different product is shown as low confidence for manual review, instead of appearing as a confident match.
Fixed
- Localized URLs are no longer skipped on multi-market sites. Source URLs that differ only by language or region from a destination were sometimes treated as “no redirect needed” and left out of the export. Every source URL that needs a redirect now appears in the results.
Migration matching refinements: loop prevention, parent-path fallback, cross-domain rebrands, automatic deduplication, gap analysis report.
New
- No Redirect Needed tier. When a source URL exists unchanged in your destination set (the URL is being kept as-is), it now gets a dedicated tier instead of a 100% self-match. These rows are excluded from the Summary and Detailed CSV exports by default, so the deployed redirect file can never contain a
/foo to /foo rule that creates an infinite loop.
- Parent path fallback tier. Deep URLs the engine cannot confidently match (discontinued products, archived posts) now fall back to their parent path when one exists in your destination set (
/products/discontinued-item-xyz to /products), preserving link equity at the category level instead of becoming a 404. Tagged at 50% so you can review or filter them.
- Cross-domain exact match tier. When the engine detects a single-domain to single-domain rebrand (
old.com to new.com with paths preserved), identical paths across the two domains are matched at 97% confidence in the exact phase. Faster and more reliable than relying on the AI phase to recognize that the path stayed the same.
- Gap Report CSV download. A new download button on the results page produces a two-section report: source URLs with no confident match (each annotated with the closest-but-too-low candidate the engine considered), and destination URLs that no source URL maps to (new content with no inbound redirect path).
Improved
- Duplicate URLs are removed automatically at upload. Deduplication collapses only exact duplicates and tracking-parameter variants of the same address, so
/foo and /foo?utm_source=newsletter become one row (a single 301 carries the query string through). Anything that is a separate resource under RFC 3986 is kept so it gets its own redirect: a different scheme (http vs https), host (example.com vs www.example.com), trailing slash (/foo vs /foo/), language path (/en/foo vs /de/foo), or file extension. The upload status reports how many duplicates were removed.
- Match breakdown shows two new tiers (Parent Fallback and No Redirect Needed) alongside the existing Exact, High, Medium, Low, Review, and No Match counts.
Fixed
- Self-redirect loops can no longer leak into exported configs. Previously, if your source and destination lists both contained
https://example.com/foo, the matcher emitted a 100% exact match from that URL to itself. Deploying that line into nginx, Apache, or Cloudflare would create an infinite redirect loop. The new No Redirect Needed tier catches this case and the CSV exporter excludes it by default.
RedirectaBot User-Agent, faster and more reliable redirect checker, cancel running and queued jobs.
New
- RedirectaBot is now a User-Agent option in the redirect checker. A branded crawler identity (
RedirectaBot/1.0 (+https://redirecta.io/bot)) for users who want our tool to identify itself in target server logs. Backed by a public info page at /bot explaining what RedirectaBot does and how to allow or block it.
- Cancel a running matching job. Stop a job mid-match from the history table or the progress screen. Files are removed cleanly.
- Cancel a queued job from the history table.
Improved
- Redirect checker is significantly faster. Typical 3-URL batches now complete in roughly half the previous time.
- Redirect checker is more reliable on sites that block bots (a common Cloudflare pattern on social platforms). The status now reflects what a real browser would see, with a hint to try a different User-Agent if needed.
- Mapping validator is friendlier. Auto-detects the delimiter and header rows, and accepts drag-and-drop directly onto the textarea.
- Minor UI polish across the app.
Fixed
- Dashboard no longer hijacks you to the last viewed job when navigating in from a public page. Refresh on the progress screen still restores where you were.
- Concurrent-job limit is now enforced.
- Admin and Dashboard buttons disappear from the navbar when your session expires.
- Project name typed after dropping the CSV is now saved.
- Elapsed time on the second job no longer carries over from the first.
- Progress bar no longer flashes 0% when returning to the matching screen.
- Minor bug fixes.
Row-count limit on CSV uploads.
Improved
- CSV uploads now have a row-count ceiling. In addition to the existing 10 MB size limit, files above 50,000 rows are rejected with a clear “split the file into smaller chunks” message. Prevents oversized jobs from queuing up and timing out part way through matching.
Cross-locale matching safety net: wrong-market matches no longer slip through as exact.
Fixed
- Cross-locale matches now surface as “no match” instead of redirecting to the wrong market. Previously, a source URL like
/de-de/products/foo could be matched at 100% to a destination like /en-nl/products/foo when no same-language alternative existed. The matcher now rejects those pairings whenever both URLs carry a locale and their primary languages differ. Same-language pairs (such as de-de to de-at, or en-us to en-gb) and locale-add or locale-drop migrations (one side has no locale prefix) are unaffected.
Cross-device job sync, smarter scoring, saved corrections, and expanded dictionaries.
New
- Cross-device job sync. Start a mapping job on one computer, pick it up on another. Active jobs and their elapsed time now stay in sync through your account, no matter where you sign in.
- Save your corrections. When you edit a match on the results page, your correction is saved against the job. Edited rows are marked so you can see at a glance what you have reviewed.
- Destination metadata crawler. Enable crawling on the destination list to fetch titles, headings, and meta descriptions. Richer context means better matches for URLs with thin or cryptic slugs.
- Expanded matching dictionaries. Added five new categories: clothing, furniture, electronics, food and drink, and home. Product and category URLs in these verticals now translate across languages with far more coverage.
- Refined scoring engine. The matcher now uses 6 composite scoring groups instead of 14 individual signals. Scores are easier to reason about, and matches are more consistent across job types.
- Job queue limit. Each account can hold up to 10 jobs. Old finished jobs can be deleted to free space.
- Edit button in job history. Re-open any completed job straight from history without navigating back through the dashboard.
Improved
- Auto-logout now works reliably. Previously, switching between the dashboard and admin panel could sign you out even within the 30-minute idle window. Sessions are now managed server-side, so the idle clock behaves the same on every page and every device.
- Results table on mobile. The View and Edit action buttons no longer sit flush against the right edge of the screen. Column widths were rebalanced so Score has more room and Details stays compact.
- Progress indicator redesign. The circular spinner on the matching screen now matches the Material 3 progress style used elsewhere in the app.
- Start New Job button. Content is now properly centered.
Fixed
- Race condition when switching between jobs rapidly could leave stale results on screen.
- Redirect checker rate limiter is now shared across requests, so the displayed remaining quota stays accurate.
Smarter matching for multilingual sites and bug fixes.
New
- Locale-aware matching. URLs with language prefixes like
/de/, /en-gb/, or /fr/ now match to the correct localized destination. Previously, a German source URL could match to an English destination if the path was identical.
- Compound word recognition. The matching engine now understands compound words common in Nordic and Germanic languages. For example, “wintermantel” is recognized as “winter” + “mantel”, improving matches for product and category URLs.
Fixed
- Redirect checker now returns correct status codes for all servers.
- Progress bar no longer jumps backwards during matching.
- Results page loads reliably after matching completes.
HTTP Status Codes glossary, favicon support, and bug fixes.
New
- HTTP Status Codes glossary: a complete reference covering all 62 HTTP response codes. Each code has its own page with an explanation, common causes, and how to fix it. Codes are grouped by class (1xx through 5xx) with easy navigation between them.
- Site icon now appears in browser tabs, bookmarks, and when saving to your phone’s home screen.
Fixed
- CSV download from the results page no longer fails with an error.
- Dark mode toggle works correctly on all pages.
Initial public release of Redirecta, the AI-powered redirect mapping tool for site migrations.
New
- AI-powered redirect matching using semantic analysis and multilingual embeddings
- Multi-signal confidence scoring for every match with full transparency
- Entity extraction for colors, sizes, product types, and categories from URL slugs
- Multilingual support: 7 dictionaries covering 1,655 terms across 12+ languages
- CSV upload with flexible column detection (URLs, titles, H1s, meta descriptions)
- Step-by-step wizard: Upload, Configure, Match, Review Results
- Configurable matching: similarity threshold, scoring weights, column selection
- Detailed results view with per-signal score breakdown
- Export results as CSV (summary or detailed)
- Job history with status tracking (up to 10 jobs per user)
- Public redirect chain checker tool
- Full documentation and getting started guide
- Blog and release notes
- Dark mode
- Responsive design for mobile, tablet, and desktop