Internal PageRank: preserving internal linking during a website migration
Internal links share a site's popularity between its pages. A migration moves them, breaks them or routes them through redirects: here is how to estimate that distribution and preserve it.
Internal links are not only for navigation: they tell Google which pages matter and share between them the popularity the site receives from outside. This is what is called the distribution of internal PageRank. A website migration directly affects this linking: a new menu, new addresses, deleted or merged pages. Without care, important pages lose their links overnight.
This article is about internal linking. The link between PageRank and website migration from the redirect point of view, and preserving external popularity signals (inbound links, listings, profiles) each have their own article.
What internal PageRank is
PageRank is the algorithm Google started with, in 1998, to rank pages by the links they receive: a page is worth more when it receives links from pages that are themselves worth a lot, and each page shares its value between the links it gives. Google has not shown a public PageRank since 2016, but links remain one of its signals.
Internal PageRank is the same logic applied to links between the pages of one site. The home page, which receives most external links, passes their value on through the menu, the footer, the breadcrumb and the links in the content. A page linked from the menu of every page receives a lot; a page five clicks from the home page, linked once, receives little; a page that no link reaches (an orphan page) receives nothing at all.
Estimating the internal PageRank distribution
The PageRank Google computes is not visible, but its distribution can be estimated from the site's link graph:
- crawl the site with a tool that records every internal link (source, target, anchor,
nofollowattribute); - read the internal popularity score some crawlers compute on this graph (Screaming Frog's Link Score, Oncrawl's Inrank, for example), or failing that the number of inbound links and the depth of each page;
- cross-check with Search Console: a page with impressions but few internal links should be linked better; a heavily linked page with no impressions uses linking for nothing.
These scores are approximations: they are useful to compare pages with each other and, above all, to compare the old site with the new one.
What a migration does to internal linking
Several changes, often decided for other reasons, alter the linking:
- the new design: a shorter menu, a lighter footer, removed "related articles" blocks take away links that were on every page;
- created, merged or deleted pages: value is shared among a different set of pages, and a page deleted without a redirect takes with it the links it gave;
- links that still target old addresses: they go through a redirect, sometimes several, instead of reaching the page directly;
- links rendered in JavaScript or behind filters, which bots follow poorly;
- recrawl time: Google has to crawl the site again to read the new linking, which takes weeks on a large site.
Preserving linking during the migration
Before: record the old site's linking
Crawl the old site and keep the export: internal links per page, depth, internal popularity score. Identify the pages that matter (traffic, conversions, external links) and note where their links come from. This is your baseline.
On staging: compare
Crawl the staging site the same way and compare page by page, using the mapping between old and new addresses. An important page that drops from fifty internal links to two, or from two clicks to five, must be linked again: menu, category, breadcrumb, links in related content. Also check that no page you keep has become an orphan.
Rewrite links to the final addresses
Redirects protect external links, which you do not control. Internal links can be fixed: every link in a piece of content or a template must point directly to the new address, not to the old one that redirects. The validated redirect table is the dictionary for this rewrite. See our guide to the redirect plan.
After go-live
Crawl the live site again: no internal link should return a 404 or go through a redirect. In Search Console, follow the crawling and indexing of important pages, and the remaining risks of redirects.
What Swan brings
Swan does not compute internal PageRank and does not analyse linking. It crawls the old site and the staging site, suggests for each old URL the new page closest in content, has you validate doubtful cases, then checks and exports the redirects. This validated table is what lets you rewrite internal links to their final addresses and compare, page by page, the linking of the old and new sites.
Is your staging site ready? Swan matches each old URL with its new page and prepares the redirects that will also serve to rewrite your internal links.
