Tool AT-W49REV. 2026-A

Redirect planner: redirect old addresses correctly at a relaunch

At a relaunch, every old address that brings visitors or links needs a permanent redirect (301 or 308) straight to the matching new page. Not a blanket redirect to the home page, and not via intermediate stops. The planner matches your old addresses to the new ones, shows the calculation for each row, finds chains and loops, and writes the finished rules for your server.

  • free
  • no sign-up
  • runs in your browser

Paste old and new addresses, take the rules with you

Planner · old addresses to new pages

The planner suggests matches and writes the rules; you still have to check them. Test the export first on a preview or with a few addresses before it goes live on the real website.

Step 1Addresses of the old and the new website

One address per line, a table (CSV, also from Excel or a crawler) or the XML sitemap. In the old list a target may already follow an address, such as /alt /neu or /alt 410; that target then applies. Up to 3,000 addresses per list.

Step 2What should count as the same in the comparison

0 to 100. Below that, an address is left without a target and waits for your decision. How the points are calculated is explained in the text below the table.

Runs in your browser. Nothing is sent or stored, not even a file you have read in.

Why old addresses need a redirect

A website does not live only on its own pages. Directories, partners, old circulars, PDF brochures, bookmarks and Google’s result lists all link to your addresses. If the addresses change at a relaunch without a redirect, all these links end up on an error page. That is no exception: of the pages that existed on the web in 2013, 38 percent had disappeared ten years later, and the larger part was missing as a single page on a domain that was otherwise still running Pew 2024. That is exactly what a relaunch without redirects looks like.

Usually the content is not gone at all; it just has a new address. For dead links in journal articles, searchers found the page in a new place or equivalent replacement in 80 percent of cases Lawrence et al. 2001. The most common error was simply 404, because files had been renamed or removed Spinellis 2003. Your customers do not search as patiently as researchers. Anyone who lands on an error page is more likely to call the next business in the results list.

What studies measure about disappeared addresses
StudyMeasuredResult
Lawrence et al. 200167,577 web addresses in computer science articles23 to 53% invalid, 80% of those recoverable
Spinellis 20034,224 web referencesHalf-life about 4 years, about 60% of errors are 404
Hennessey and Ge 201314,489 web pages from abstractsMedian lifetime 9.3 years
Zittrain et al. 2014Law journals and the US Supreme Courtover 70% and 50% respectively no longer lead to the source
Klein et al. 2014over 1 million references in 3.5 million articlesone in five articles affected
Jones et al. 2016Content behind references that still workchanged in over 75%
Zittrain et al. 2021Links in the New York Times 1996 to 2019Share of dead links grows almost linearly with age
Pew 2024almost 1 million pages from Common Crawl38% of pages from 2013 gone after 10 years

How long should the redirects stay in place? Google says “as long as possible, usually at least one year” so that all signals move over to the new addresses (as of 20 August 2026). The links out there live longer: web pages cited in journal articles lasted a median of 9.3 years Hennessey and Ge 2013, and the share of dead links in newspaper articles keeps growing with every year of publication Zittrain et al. 2021. That is why we leave redirects in place for as long as the old domain is in your hands.

Arriving somewhere is not enough, by the way. For over three quarters of the references in journal articles that could still be reached, the address by then led to different content Jones et al. 2016. For your customers that means: anyone who searched for “server maintenance” wants to land on the server maintenance page, not start again from the home page.

301, 302, 307 or 308: which code for what

Which number the server sends is governed by the HTTP standard RFC 9110 (June 2022). 301 and 308 say: this page has moved permanently, future links should use the new address. 302 and 307 say: only temporary, please keep using the old address. The difference within each pair concerns forms. With 301 and 302 the browser may, for historical reasons, turn a POST into a GET; with 307 and 308 it may not.

Status codes for redirects according to RFC 9110 and Google
CodeMeaningMethod (POST)At GooglePurpose
301Moved Permanently, permanentmay change from POST to GET for historical reasonsstrong signal: the new address becomes the canonical onestandard for a relaunch and a domain move
308Permanent Redirect, permanentstays the same (POST stays POST)as 301form and API addresses that receive POST
302Found, temporarymay change from POST to GET for historical reasonsweak signal: the old address usually stays in the indexmaintenance, time-limited campaigns, printed short links
307Temporary Redirect, temporarystays the same (POST stays POST)as 302temporary, when POST has to be preserved
410Gone, removedno redirectpage drops out of the index, as with 404content removed on purpose, no successor

Sources: RFC 9110, sections 15.4.2 to 15.4.9 and 15.5.11; Google “Redirects and Google Search”, as of 14 April 2026; Google “HTTP status codes”, as of 4 February 2026.

Google treats 301 and 308 equally as a strong signal that the new address belongs in the search results. 302 and 307 are a weak signal, and the old address then usually stays in the index. For a relaunch, 301 is therefore the standard choice; use 308 for old addresses that receive data from forms or APIs. If content has been removed on purpose and has no successor, the server is better off answering with 410. Google treats that like 404 and takes the address out of the index; for people and other programs, 410 also says that this is intended.

How the planner matches old addresses to new ones

First it brings all addresses into a uniform form, as RFC 3986 provides: scheme and domain in lower case, %-sequences with upper-case hex digits, segments such as /./ resolved. Then you decide what else should count as the same. The same page is often found under several addresses, through aliases, redirects and start files Bar-Yossef et al. 2009. For the comparison, the planner aligns such variants. Each one is still redirected individually, because Apache distinguishes /Kontakt from /kontakt.

Variants of the same address
VariantOld addresstreated as the same asSetting
Case/Leistungen/Webdesign.html/leistungen/webdesign.htmlUpper and lower case
Slash/team//teamTrailing slash
Start file/ordner/index.php/ordner/Index files
Tracking/kontakt?utm_source=flyer/kontaktTracking parameters
Order/shop?b=2&a=1/shop?a=1&b=2always
Encoding/%7Eteam/~teamalways (RFC 3986)

The score: words, name and path

If an old address, after alignment, appears in exactly that form in the new list, the matter is clear: same path, 100 points. Otherwise the planner calculates a score from 0 to 100 for each pair, made up of three parts.

  • Words (50%). Path and query values are split into words at special characters and at changes between letters and digits, endings such as .html are dropped, and umlauts are folded: ü counts as ue and as u. The address alone reveals a surprising amount. A URL split up in this way classified pages about three quarters as well as the full page text Kan 2004. What is counted is the share of shared words (Jaccard). A word that is the beginning of another (leistung, leistungen) also counts as the same, as does one with a typo.
  • Name (30%). The last segment without separators, compared by edit distance: the smallest number of characters you have to replace, delete or insert Wagner and Fischer 1974. Divided by the length of the longer name, this gives 0 to 100 percent, so short and long names remain comparable Navarro 2001.
  • Path (20%). The same for the whole path. A folder that has moved therefore costs fewer points than a renamed name.

The mix has a reason. Character comparisons catch typos and renamings but fail with reordered words; word comparisons make up for exactly that Elmagarmid et al. 2007. From 50 points the planner matches automatically. If the runner-up is less than 5 points behind, it marks the row as uncertain. The weights and the threshold are values from our experience, calibrated on examples, and not a standard. That is why the calculation is shown in every row.

Scores from the planner’s example
Old addressNew pageWordsNamePathPoints
/leistungen/webdesign.html/webdesign50 %100 %45 %64
/kontackt.php/kontakt100 %88 %88 %94
/index.php?seite=impressum/impressum100 %100 %0 %80
/aktuelles/2019/sommerfest-2019/blog/sommerfest-201950 %100 %57 %66
/ueber-uns.html/über-uns100 %100 %100 %100
/leistungen/seo-optimierung.html/suchmaschinenoptimierung (by hand)0 %54 %48 %26

Example /leistungen/webdesign.html → /webdesign: 50% × 0.5 + 100% × 0.3 + 45% × 0.2 = 64 points. The tracking parameter utm_source on /kontackt.php is removed before the comparison.

The last row shows the limit: “seo-optimierung” and “suchmaschinenoptimierung” mean the same thing but share no word and only a few letters. With 26 points the address stays without a target, and you decide by hand in the table. That is quick, because the input field offers all new addresses to choose from.

Finding chains, loops and catch-all targets

CORRECT: ONE HOP/leistungen.html301 DIRECT/webdesignCHAIN: THREE HOPS/leistungen.html/leistungen/angebot/webdesignLOOP: NO END/kontakt/kontakt/BACK TO /kontaktCORRECT: ONE HOP/leistungen.html301 DIRECT/webdesignCHAIN: THREE HOPS/leistungen.html/leistungen/angebot/webdesignLOOP: NO END/kontakt/kontakt/BACK TO /kontakt
FIG. 1 · DIRECT, CHAIN, LOOPGOLD = THE PAGE WHERE THE VISITOR SHOULD ARRIVE
  • Chains. A redirects to B, B to C, often after the second relaunch. Googlebot follows up to 10 hops, but Google advises redirecting straight to the final target. Every hop costs loading time. The planner resolves chains and writes only the final target into the export.
  • Loops and self-references. A redirects to B and B back to A, or an address to itself. The browser gives up with “too many redirects”, and Google does not index the page. Such rows are missing from the export until a target is changed.
  • Only the spelling changes. /kontakt to /kontakt/ is fine as long as the new website does not itself redirect /kontakt back to /kontakt/. That creates a loop that you only see on the server. The planner points out such pairs.
  • Catch-all target: home page. Sending many old addresses to the home page seems convenient. Google explicitly advises against it and may treat it as a soft 404, in other words like a hidden error page (as of 20 August 2026). Such substitute pages already made up around 16 percent of dead links in 2004 Bar-Yossef et al. 2004. From 3 addresses pointing to the home page, the planner warns you.
  • Without target. If the planner finds nothing, look for the successor page using the title of the old page. Via the title, disappeared pages were found as the first result of a search engine in over 60 percent of cases Klein and Nelson 2010. If there really is no successor, 410 is more honest than the home page.
  • Target without a page, live source. A target entered by hand that is not in the new list probably ends on an error page. And if an old address is also in the new list, the redirect would hide a page that still exists.

Rules for Apache, nginx and Caddy

Every server has its own syntax and its own pitfalls. The export takes them into account. We checked the rules against the documentation and, where that is silent, against the servers’ source code: how Apache and nginx read quotation marks and backslashes in the configuration, and what mod_rewrite evaluates in the substitution. Our test reads every export back the way the server reads it (as of 14 November 2026).

How the export writes for each server
ServerWhere toSimple pathWith query stringWhat the export takes care of
Apache 2.4.htaccess in the root directory, at the very topRedirect 301RewriteCond %{QUERY_STRING} + RewriteRule [R=301,L,NE,QSD]Redirect also redirects everything below the address. If the target lies below the old address (/leistungen → /leistungen/uebersicht), a loop would result; the export then uses mod_rewrite with an exact match. Without NE, mod_rewrite would encode %-sequences twice.
nginxmap in the http block, if and return in the server blockmap $urimap $request_uri with regexmap compares strings without regard to case. If /Kontakt redirects to /kontakt, the rule would also match /kontakt itself; for that the export writes a regex. A $ in the target becomes %24, otherwise nginx would take it for a variable.
Caddy 2Site block of the old domainredir /alt /neu{?query} 301named matcher with path and queryredir without a code redirects only temporarily (302); the export always writes out the code. path compares without regard to case; for collisions the export uses path_regexp.
CSVSpreadsheet program, redirect extension in the CMSone row per old address with target, status, type and pointsSemicolon, quotation marks per RFC 4180, no cell content begins with =, +, - or @ (no formulas).

Addresses such as /index.php?seite=kontakt are common on older websites. For these the planner writes rules that check each parameter individually. Order and additional parameters such as tracking do not matter, but /index.php?seite=kontakt2 does not match the rule. With such rules the old query string is dropped at the target; with all others the server passes it on.

The example as a finished export

Apache (.htaccess)
# Weiterleitungen aus dem Weiterleitungs-Planer, https://aureum-tech.com/tools/weiterleitungs-planer
# Stand der Regeln: 14. November 2026. 8 Regeln, Status 301.
# Vor dem Einsatz testen. Weiterleitungen dauerhaft behalten, mindestens 1 Jahr (Google).
# Ganz oben in die .htaccess im Hauptverzeichnis der alten Website, vor die Regeln des CMS.

# mod_alias: Redirect gilt für den Pfad und alles darunter und hängt den Query-String an.
Redirect 301 "/aktuelles/2019/sommerfest-2019" "/blog/sommerfest-2019"
Redirect 301 "/leistungen/webdesign.html" "/webdesign"
Redirect 301 "/Leistungen/Webdesign.html" "/webdesign"
Redirect 301 "/ueber-uns.html" "/%C3%BCber-uns"
Redirect 301 "/kontackt.php" "/kontakt"
Redirect 301 "/referenzen/" "/referenzen"

# mod_rewrite: Adressen mit Query-String und Pfade, unter denen Seiten weiterleben (exakter Vergleich).
RewriteEngine On
RewriteCond %{QUERY_STRING} "(^|&)seite=impressum(&|$)"
RewriteRule "^index\.php$" "/impressum" [R=301,L,NE,QSD]
RewriteRule "^index\.php$" "/" [R=301,L,NE]

# Ohne Ziel (bitte zuordnen oder 410 eintragen):
#   /leistungen/seo-optimierung.html
#   /team.html
nginx
# Weiterleitungen aus dem Weiterleitungs-Planer, https://aureum-tech.com/tools/weiterleitungs-planer
# Stand der Regeln: 14. November 2026. 8 Regeln, Status 301.
# Vor dem Einsatz testen. Weiterleitungen dauerhaft behalten, mindestens 1 Jahr (Google).

# Teil 1 in den http-Block, z. B. /etc/nginx/conf.d/weiterleitungen.conf.
# Zeichenketten vergleicht map ohne Groß-/Kleinschreibung, "~…" mit.

map $request_uri $weiterleitung_query {
    default "";
    "~^/index\.php\?(?=(?:.*&)?seite=impressum(?:&|$))" "/impressum";
}

map $uri $weiterleitung_ziel {
    default "";
    "/aktuelles/2019/sommerfest-2019" "/blog/sommerfest-2019";
    "/leistungen/webdesign.html" "/webdesign";
    "/ueber-uns.html" "/%C3%BCber-uns";
    "/kontackt.php" "/kontakt";
    "/referenzen/" "/referenzen";
    "/index.php" "/";
}

# Teil 2 in den server-Block der alten Domain:
if ($weiterleitung_query) { return 301 $weiterleitung_query; }
if ($weiterleitung_ziel) { return 301 $weiterleitung_ziel$is_args$args; }

# Meldet nginx „could not build map_hash", hilft im http-Block: map_hash_bucket_size 256;

# Ohne Ziel (bitte zuordnen oder 410 eintragen):
#   /leistungen/seo-optimierung.html
#   /team.html
Caddy
# Weiterleitungen aus dem Weiterleitungs-Planer, https://aureum-tech.com/tools/weiterleitungs-planer
# Stand der Regeln: 14. November 2026. 8 Regeln, Status 301.
# Vor dem Einsatz testen. Weiterleitungen dauerhaft behalten, mindestens 1 Jahr (Google).
# In den Site-Block der alten Domain, z. B. example.com { … }.
# path vergleicht ohne Groß-/Kleinschreibung, path_regexp mit.

@weiterleitung_1 {
    path /index.php
    query seite=impressum
}
redir @weiterleitung_1 /impressum 301
redir /aktuelles/2019/sommerfest-2019 /blog/sommerfest-2019{?query} 301
redir /leistungen/webdesign.html /webdesign{?query} 301
redir /ueber-uns.html /%C3%BCber-uns{?query} 301
redir /kontackt.php /kontakt{?query} 301
redir /referenzen/ /referenzen{?query} 301
@weiterleitung_2 path /index.php
redir @weiterleitung_2 /{?query} 301

# Ohne Ziel (bitte zuordnen oder 410 eintragen):
#   /leistungen/seo-optimierung.html
#   /team.html

How to plan the redirects for a relaunch

  1. Secure the inventory before you switch over. Old sitemap (usually /sitemap.xml), Google Search Console (Performance and Links) and visitor statistics. A crawl of the old website finds the rest. After the switch this list is hard to get.
  2. Match. Put the old and the new list into the planner, go through the suggestions, and decide uncertain rows and addresses without a target by hand.
  3. Decide what goes. Content without a successor gets 410. Pages with links or visitors are often worth a new page rather than a farewell.
  4. Test on the preview. Load the rules and call up every old address: a single hop, status 200 at the end. For many addresses a crawler with the old list helps.
  5. Switch over and tidy up. Internal links, sitemap and canonical tags point to the new addresses. For a domain change, report the move in Search Console with the Change of Address tool.
  6. Watch. In the first weeks, read the error reports in Search Console and close gaps. The redirects stay in place.

Where the planner stops

  • By address only. It does not fetch any page and compares neither content nor titles. Whether a target really responds, only the test on the server shows.
  • Server peculiarities. Existing rules, the CMS and settings for trailing slash, www or https can interact with the export. That is why you should test on a preview first.
  • Score as a guide. The weights and the threshold are values from experience. It does not recognise words with the same meaning but different letters.
  • Quantities. In the browser up to 3,000 addresses per list, via the interface for AI assistants up to 500. Split larger websites by section.

How we approach a relaunch so that no page gets lost is described under Web design: replacing the old website. If pages or visitors are already missing after a rebuild, carry on with Homepage is not found, and the guide Website not found on Google helps you look for the cause. More tools around findability are listed under SEO tools. An overall verdict on your website, from technology to content, is provided by SpotLens.

Using this tool from an AI assistant

Assistants such as Claude or ChatGPT that speak the Model Context Protocol (MCP) can call this page’s calculation directly instead of rebuilding it. The server is public, needs no sign-in and stores nothing. The answer always includes the link to this page, the date of the data and the limits of the result.

Endpoint
https://aureum-tech.com/mcp
Tool
werkzeug_weiterleitungs_planer
Transport
Streamable HTTP, read-only
Inputs
NameRequiredMeaning
alte_urlsyesAddresses of the old website. A row may additionally contain a target or 410 (“/alt /neu”, “/alt;/neu”, “/alt 410”), in which case that applies: as a list of addresses or as text (one address per line, CSV with header row or a pasted XML sitemap). Absolute addresses (https://…) or paths (/…). At most 500 addresses and 200,000 characters; please split larger lists by section or use the page in the browser.
neue_urlsnoAddresses of the new website (e.g. its sitemap). May be omitted if the old list already contains targets: as a list of addresses or as text (one address per line, CSV with header row or a pasted XML sitemap). Absolute addresses (https://…) or paths (/…). At most 500 addresses and 200,000 characters; please split larger lists by section or use the page in the browser.
zuordnungennoOptional: matches made by hand that override the automatic matching.
formatnoExport format: apache (.htaccess), nginx, caddy or csv. apache if not specified.
statusnoStatus code of the redirects: 301 or 308 (permanent), 302 or 307 (temporary). 301 if not specified.
schwellenoMinimum score (0–100) for an automatic match. 50 if not specified.
gross_klein_ignorierennoIgnore upper and lower case in the comparison. Yes if not specified.
slash_ignorierennoIgnore the trailing slash in the comparison. Yes if not specified.
index_ignorierennoIgnore index.html, index.php, default.aspx in the comparison. Yes if not specified.
tracking_entfernennoRemove utm_*, gclid, fbclid and similar parameters from old addresses. Yes if not specified.
ohne_ziel_als_410noWrite addresses without a target in the export as 410 (removed) instead of just a comment. No if not specified.
ziele_absolutnoAlways write targets with the domain. If not specified, only when the domain changes.

Questions where an assistant should call it

  • “We are doing a relaunch: which old address has to redirect to which new page? Here are both sitemaps.”
  • “Write me the .htaccess redirects for these old and new URLs.”
  • “Does our redirect plan contain chains or loops?”
  • “How do I redirect /index.php?seite=kontakt to /kontakt in nginx?”
  • “Should I use 301 or 308 for a domain move, and what do I do with pages that no longer exist?”

How the connection works and what MCP can do in your own business is explained on the page about MCP integrations.

Common questions about redirects at a relaunch

Which redirect for a relaunch: 301 or 308?

Both are permanent, and Google treats both as an equally strong signal that the new address belongs in the search results. The difference concerns forms: under RFC 9110 a browser may turn a POST into a GET after a 301, but not after a 308. For ordinary pages use 301; for addresses that receive form submissions or API requests, use 308. 302 and 307 are temporary; in that case the old address usually stays in the index.

How long do I have to keep redirects after a relaunch?

Google recommends as long as possible, usually at least 1 year, so that all signals move over to the new addresses (as of 20 August 2026). But links on other websites, in PDFs and in bookmarks last much longer: web pages cited in journal articles lived a median of 9.3 years. That is why we leave redirects in place for as long as the old domain is in your hands.

Can I simply redirect old pages to the home page?

Better not. Google explicitly advises against sending many old addresses to an unsuitable target such as the home page, and may treat that as a soft 404, in other words like an error page. Visitors do not land on what they were looking for, either. Redirect each address to the page that matches its content; if there is none, the server answers with 410. The planner warns you from 3 addresses pointing to the home page.

What is a redirect chain, and why is it harmful?

A chain arises when A redirects to B and B to C, for example after two relaunches. Googlebot follows up to 10 hops, but Google advises redirecting straight to the final target. Every hop costs loading time, and if the chain runs in a circle, the browser gives up. The planner resolves chains and writes only the final target into the export.

How do I redirect addresses with parameters such as index.php?id=5?

In Redirect, Apache compares only the path; the query string is seen only by mod_rewrite through RewriteCond %{QUERY_STRING}. nginx needs $request_uri for this, Caddy the query matcher. The planner writes such rules itself: each parameter is checked individually, order and additional parameters do not matter, and the old query string is dropped at the target.

What do I do with pages that no longer exist after the relaunch?

If there is a new page with matching content, redirect to it. If the content has been removed on purpose, the server answers with 410 (Gone). Under RFC 9110 that means: removed deliberately and permanently. Google treats 410 like 404 and takes the address out of the index. In the planner you enter 410 as the target, or have all addresses without a target written out as 410.

Where do I get the list of my old addresses?

From the old sitemap (usually /sitemap.xml), from Google Search Console (Performance and Links, where the pages with clicks and incoming links are listed) and from your visitor statistics. A crawl of the old website finds the rest. It is important to save the list before you switch over; afterwards the old website can no longer be reached.

Where does the redirect planner stop?

It matches by address and does not fetch any page: it does not compare content or titles, and only a test will show whether the targets really respond on the server. The score is a guide. We take on the relaunch itself, the redirects and the check afterwards for businesses in Rodewisch and the Vogtland.

What the tool is based on

The guide values and notes above are based on these 15 scientific papers. Each one has been checked via its DOI or the publisher’s page (most recently on 14 November 2026): it exists, it has not been retracted, and its findings still hold for this question today.

  1. Lawrence, S.; Pennock, D. M.; Flake, G. W. et al. (2001): Persistence of Web references in scientific research. Computer (IEEE) 34 (3), 26–31

    Finding Of 67,577 web addresses in computer science journal articles, 23 to 53 percent were invalid depending on the year. For 80 percent of the dead links, searchers found the page at a new address or equivalent information, mostly via a search engine.

    In the tool The core of the tool: most dead links point to content that has merely moved. A redirect saves visitors the detour via search.

  2. Spinellis, D. (2003): The decay and failures of web references. Communications of the ACM 46 (1), 71–77

    Finding For 4,224 web references the half-life was about four years. The most common error was 404 (around 60 percent), mostly because files had been renamed or removed; references to files decayed more often than those to directories.

    In the tool That is why the tool compares names without endings such as .html and .php and treats the renamed file as the normal case of a relaunch.

  3. Hennessey, J.; Ge, S. X. (2013): A cross disciplinary study of link decay and the effectiveness of mitigation techniques. BMC Bioinformatics 14 (Suppl. 14), S5

    Finding 14,489 web pages from abstracts in the Web of Science (1996 to 2010) lived a median of 9.3 years. Web archives captured addresses with few path levels better than deep ones.

    In the tool Supports the recommendation to keep redirects for many years rather than just the one year Google gives as a minimum, and not to forget deep addresses.

  4. Klein, M.; Van de Sompel, H.; Sanderson, R. et al. (2014): Scholarly Context Not Found: One in Five Articles Suffers from Reference Rot. PLoS ONE 9 (12), e115253

    Finding Over a million web references from 3.5 million journal articles (1997 to 2012): one in five articles suffers from reference rot, meaning dead links or changed content; among articles with web references the figure is seven in ten.

    In the tool Shows in the explanatory section how many references to an old address are still in other people’s documents that nobody changes any more.

  5. Jones, S. M.; Van de Sompel, H.; Shankar, H. et al. (2016): Scholarly Context Adrift: Three out of Four URI References Lead to Changed Content. PLOS ONE 11 (12), e0167475

    Finding For over 75 percent of the references, the content behind the address had changed since it was cited (content drift), measured with established similarity measures for text.

    In the tool That is why it is not enough for an old address to arrive somewhere: the planner looks for the page that matches in content and warns about catch-all targets such as the home page.

  6. Zittrain, J.; Albert, K.; Lessig, L. (2014): Perma: Scoping and Addressing the Problem of Link and Reference Rot in Legal Citations. Legal Information Management 14 (2), 88–99

    Finding More than 70 percent of the addresses in the Harvard Law Review and other journals, and 50 percent of the addresses in US Supreme Court opinions, no longer led to the cited information.

    In the tool Shows in the explanatory section that dead references are the rule, not the exception, even in carefully maintained sources.

  7. Zittrain, J.; Bowers, J.; Stanton, C. (2021): The Paper of Record Meets an Ephemeral Web: An Examination of Linkrot and Content Drift within The New York Times. SSRN Electronic Journal (Harvard Law School)

    Finding In New York Times articles from 1996 to 2019, the share of dead links rose almost linearly with age. More than half of the articles with at least one link contained a dead one.

    In the tool Explains why redirects should not be switched off after a year: otherwise every year’s worth of old links breaks a little further.

  8. Chapekis, A.; Bestvater, S.; Remy, E.; Rivero, G. (2024): When Online Content Disappears. Pew Research Center, Data Labs, 17 May 2024

    Finding Of almost a million pages from Common Crawl, 38 percent of the pages from 2013 could no longer be reached in 2023. Of all pages checked, 16 percent were missing individually on an otherwise reachable domain, 9 percent along with the whole domain.

    In the tool The most common case is the disappeared single page on a living website. That is exactly what a relaunch without redirects produces.

  9. Bar-Yossef, Z.; Broder, A. Z.; Kumar, R.; Tomkins, A. (2004): Sic transit gloria telae: towards an understanding of the web's decay. Proceedings of the 13th International Conference on World Wide Web (WWW 2004), 328–337

    Finding Many servers do not report a missing page with 404 but deliver a substitute page or redirect to the home page. These “soft 404s” made up around 16 percent of the dead links in the sample and are hard for machines to recognise.

    In the tool That is why the planner warns when several old addresses point wholesale to the home page, and offers 410 for removed content instead of a redirect into nothing.

  10. Bar-Yossef, Z.; Keidar, I.; Schonfeld, U. (2009): Do not crawl in the DUST: Different URLs with Similar Text. ACM Transactions on the Web 3 (1), 1–31

    Finding The same page can often be reached under several addresses, through aliases, redirects and dynamically generated pages. The rules behind this can be derived from address lists and logs without fetching the pages.

    In the tool Basis of the variant check: upper and lower case, trailing slash, index.html and tracking parameters are aligned for the comparison, and each is redirected separately.

  11. Kan, M.-Y. (2004): Web page classification without the web page. Proceedings of the 13th International World Wide Web Conference, Alternate Track Papers & Posters, 262–263

    Finding The address alone, split into words at special characters and at changes between upper-case and lower-case letters and digits, classified pages about three quarters as well as the page text and better than titles or link texts.

    In the tool The tool splits paths into words in the same way and matches without fetching a page.

  12. Klein, M.; Nelson, M. L. (2010): Evaluating methods to rediscover missing web pages from the web infrastructure. Proceedings of the 10th Annual Joint Conference on Digital Libraries (JCDL 2010), 59–68

    Finding Disappeared pages were found as the first result of a search engine in over 60 percent of cases via their title or a lexical signature; the title first and then the signature brought it to over 75 percent.

    In the tool For addresses without a match, the planner advises looking for the successor page via the title of the old page rather than sending it to the home page.

  13. Elmagarmid, A. K.; Ipeirotis, P. G.; Verykios, V. S. (2007): Duplicate Record Detection: A Survey. IEEE Transactions on Knowledge and Data Engineering 19 (1), 1–16

    Finding Character-based measures such as edit distance catch typos but fail with reordered words; word-based measures make up for that. Following Monge and Elkan, two words also count as the same if one is the beginning of the other.

    In the tool That is why the score mixes a word comparison (with tolerance for prefixes and typos) with the edit distance of name and path.

  14. Wagner, R. A.; Fischer, M. J. (1974): The String-to-String Correction Problem. Journal of the ACM 21 (1), 168–173

    Finding The smallest number of replacements, deletions and insertions between two strings can be calculated with a method whose running time grows with the product of the two lengths.

    In the tool The tool calculates the edit distance for name and path with exactly this method.

  15. Navarro, G. (2001): A guided tour to approximate string matching. ACM Computing Surveys 33 (1), 31–88

    Finding An overview of error-tolerant string search with a focus on edit distance: its statistical behaviour, its history and a comparison of the algorithms.

    In the tool Justifies edit distance as a measure for typos and renamings. The tool divides it by the length of the longer text so that short and long paths remain comparable.

Standards, authorities and manufacturer data

Mon–Fri 8:00–17:00

Have you got your plan? Send us the copied summary or the CSV. We check the matching, set up the redirects on the server and test them after the switch, in Rodewisch and throughout the Vogtland.

+49 3744 4399760[email protected]

Mon–Fri 8:00–17:00 · Straße der Jugend 5, 08228 Rodewisch

Discuss your relaunch
+49 3744 4399760Mon–Fri 8:00–17:00
CallCallback

PARTS LIST · SHEET 1/1

What this page is made of

Values measured at build time. With JavaScript, your browser measures them again live.

ITEMPARTTYPESIZE
01index.htmlDocument≈ 29 kB
02styles.cssStylesheet≈ 20 kB
03scripts.jsScript≈ 13 kB
04archivo-variable.woff2Font≈ 58 kB
05ibm-plex-mono.woff2Font≈ 30 kB
06favicon.svgImage< 1 kB

Advertising or analytics cookies: none. Without your consent, the page loads nothing from third-party servers apart from Cloudflare, our delivery service. Fonts, scripts and images are hosted by us. That is how we build your IT, too.

OUTSIDE OFFICE HOURS

Feel free to call anyway

Our regular office hours are Monday to Friday, 8:00–17:00.

In the evening and at weekends the number is forwarded to us, and often someone will pick up. We cannot promise that, though.

If nobody answers, leave a short message on the voicemail: your name, your number and what it is about. We will call back as soon as we can, on the next working day at the latest.

The call costs nothing. If someone has to start work straight away outside office hours, the surcharge in the price list applies.

Call now+49 3744 4399760

Prefer to write? Use the contact form or e-mail [email protected]. We will read it on the next working day.

DIAGNOSIS · AT-K2

What do you need help with?

↵ SELECTESC CLOSECALL: +49 3744 4399760

Custom settings

Which services may load?

Strictly necessary

The choice you make here and your settings for colour scheme and motion. These are kept as small entries on your device and are never sent to a server. They are needed for the site to work and therefore cannot be switched off.