Internal linking at scale is the ranking lever you actually control
Backlinks depend on other people. Content quality depends on time and budget. Algorithm updates depend on Google's mood. Internal links depend on nobody but you. They are the one major ranking input you control completely, instantly, and for free, which makes it genuinely strange how few sites manage them deliberately.
On a ten-page site, internal linking takes care of itself. At ten thousand pages, it becomes an engineering problem, and the sites that treat it like one consistently outperform the ones that sprinkle links wherever a writer happened to think of them. Here is how to approach it.
Why internal links carry so much weight
Internal links do three jobs at once. They let crawlers discover pages (a page with no links pointing at it may as well not exist). They distribute authority: the equity your domain has earned flows through internal links, and where you point them decides which pages receive it. And they describe pages: the anchor text of links pointing at a page is one of the clearest statements of what that page is about that you can give a search engine.
Every one of those jobs is under your control. No outreach, no budget approval, no waiting. That is the whole pitch.
Architecture: hubs and clusters, not spaghetti
At scale you need a structure, and the one that keeps working is the hub and cluster model. A hub page targets a broad topic. Cluster pages target the specific subtopics underneath it. Every cluster page links up to its hub, the hub links down to every cluster page, and cluster pages link sideways to their genuinely related siblings.
This does two things. It concentrates authority on the hub, which is usually the page targeting the most valuable head term. And it draws a topical boundary that helps search engines (and by now, LLM crawlers building their own maps of your site) understand that these thirty pages form one coherent area of expertise rather than thirty disconnected posts.
The related discipline is click depth. Anything you want to rank should be reachable within three or four clicks of the homepage. Most large-site audits turn up valuable pages sitting seven or eight clicks deep, buried under pagination and date archives, and their rankings reflect it. Depth is a statement of importance whether you intend it or not.
Anchor text discipline
Anchor text is where internal linking gives you a freedom external links never will: you write every word of it. Use that. "Click here" and "read more" throw the descriptive job away entirely. An anchor should say what the destination page is about, in natural language, with reasonable variation.
Descriptive beats generic: "our guide to composting in small apartments" beats "learn more".
Vary the phrasing. Two hundred internal links with the identical exact-match anchor reads as templated, because it is. Mix the head term with natural variants.
Match the anchor to the destination's target query, not to whatever sentence happened to be convenient.
One target per anchor concept: if two different pages keep receiving the same anchor text, you have just told Google they compete, and you will not like how it resolves the tie.
Orphans: the pages nobody invited
An orphan page has zero internal links pointing at it. It might be in your sitemap, it might even be indexed, but your site's structure says it does not matter, and rankings usually agree. Orphans accumulate silently: products that fell out of category listings, posts whose tags were deleted, landing pages built for campaigns and forgotten.
Finding them requires comparing two lists: pages that exist (sitemap, CMS export, or server logs) versus pages discovered by crawling links from the homepage. Screaming Frog does this directly when you connect it to your sitemaps and Search Console. Every URL that exists but was never reached by the crawl is an orphan. Then decide: pages worth keeping get linked from somewhere relevant, and pages not worth keeping get retired properly instead of haunting the index.
Navigation links vs contextual links
Not all internal links are equal. Links in your header, footer, and sidebar are template links: they appear on every page, which makes them powerful for crawling and click depth but weak as relevance signals, because a link that appears everywhere says nothing specific about anything. Contextual links, placed in the body of a page and surrounded by relevant text, are the opposite: fewer, but far more meaningful.
The practical consequence: you cannot fix a page's rankings by stuffing it into the footer. Plenty of sites run sixty-link mega-footers that exist purely for SEO, and the pattern gives nothing to the linked pages. Navigation should serve users finding things. Relevance should come from editorial, in-content links written by someone who understands both pages.
This is also the lens for template links versus editorial links generally. A "related products" block generated by a template is a template link, even though it appears in the body. It helps discovery. It does not carry the weight of a sentence in the product description that links to the buying guide with a descriptive anchor. You need both layers, but only one of them is a relevance signal you author.
Pagination effects
Paginated listings are where internal link equity goes to leak. A product on page 14 of a category receives a trickle of authority through thirteen weak hops. A blog post from two years ago sits behind twenty clicks of "older posts". If your only path to a page runs through deep pagination, you have effectively soft-orphaned it.
Mitigations that work: keep category pages linking to their most important items directly (best sellers, featured picks) regardless of pagination position, link generously from newer content to older evergreen pieces, and make sure paginated pages themselves remain crawlable with plain anchor tag links rather than JavaScript-only load-more buttons. And do not noindex deep pagination pages casually; Google has indicated that pages kept noindexed long-term tend to be crawled less, which chokes the discovery path to everything behind them.
Auditing at scale with a crawler
Past a few hundred pages, internal linking cannot be audited by hand, and it should not be audited by vibes. A crawler gives you the numbers. A standard pass in Screaming Frog:
Crawl the full site and export inlink counts per URL. Sort ascending: the bottom of that list is your neglected inventory.
Check crawl depth per URL. Anything important sitting deeper than four clicks gets a path shortened.
Export all anchor text for your priority pages. You are looking for generic anchors, duplicate anchors pointing at different URLs, and anchors that describe the wrong thing.
Cross-reference with sitemaps and GSC to surface orphans.
Find links pointing at redirects and 404s, and repoint them at final destinations. At scale, redirect chains quietly tax every hop.
Rerun quarterly. Internal linking decays constantly because every published page, retired product, and template change moves the graph.
Point the graph at the money
Here is the strategic layer most audits skip: internal links are a budget, and most sites spend theirs evenly across pages that are not evenly valuable. Your blog posts, guides, and informational pages accumulate the most links naturally. Your commercial pages, the ones that pay for everything, often accumulate the fewest.
Deliberately reverse that. Identify your money pages, then make sure your strongest and most-linked content links to them contextually, with meaningful anchors. Every informational piece in a cluster should hand its authority onward to the commercial page it supports. This is the cheapest revenue-relevant SEO work that exists, and in most audits it is simply not being done.
The trouble with automated related-links widgets
Every large site eventually reaches for automation: a widget that computes "related articles" and injects five links under every post. Tempting, but be honest about what it produces. The matching is usually shallow (shared tags, title similarity), the anchors are just the destination titles, and the links change every time the algorithm recalculates, so no page receives a stable signal. Worst case, the widget links every page to your most popular pages, making strong pages stronger and leaving the long tail exactly as orphaned as before.
Automation is fine as a discovery floor. It is not a substitute for the deliberate layer: hub-and-cluster wiring, hand-written contextual links on priority pages, and anchors an actual human chose. A workable rule for the pages that matter: automate the bottom of the page, author the middle of it.
Control is the whole appeal here. Nobody can build these links for you, and nobody can take them away. Spend the control deliberately.