Internal Linking When Your Site Has Almost No Authority

Here is the short answer, because it is the part people skip. On a site with a strong backlink profile, internal linking is a refinement — you are shaping the flow of authority you already have. On a site with almost no external links, it is not a refinement. It is the mechanism. It is the only lever over discovery, crawl priority and topical structure that you can pull entirely by yourself, without persuading a single other website to do anything. Most site owners in that position spend their effort chasing links they will not get, and leave the lever they already hold untouched.

How a hundred posts become a hundred orphans

An orphan page is a page with no internal links pointing at it. The textbook definition implies something unusual happened — a page published outside the navigation, a broken template. In practice the most common orphan is created by a completely normal setup working exactly as designed.

The setup: you publish blog posts. Each appears on /blog, which is paginated ten to a page, and new posts push old ones back. After a hundred posts, post seventy-three lives on page eight of the index, and page eight is linked from page seven, which is linked from page six. Nothing else on the site links to post seventy-three at all — not the homepage, not a service page, not another post.

That post is technically reachable, so it is not technically an orphan. But functionally it has one internal inbound link, from an archive page that itself has one, from an archive page that itself has one. Every unit of authority the domain possesses arrives at the homepage and then has to trickle through eight sequential hops to reach it. Practically nothing gets there.

Two things follow. The homepage accumulates whatever external authority the site earns and passes almost none of it onward in a targeted way. And crawl attention concentrates on the shallow pages — usually the commercial pages that never change — rather than the deep ones, which are usually the content you wrote to rank.

How do you diagnose an internal linking problem?

This is measurable, not a matter of opinion. Five checks, in order of how much they tell you per minute spent:

The output of those five checks is a short list: URLs with real impressions, low internal inbound link counts and high click depth. That list is the work.

Why click depth matters more when authority is scarce

Click depth is the number of clicks required to reach a page from the homepage using the shortest available path. It is a proxy for two things: how often a crawler reaches the page, and how much of the site's accumulated authority survives the journey.

On a site with hundreds of referring domains spread across dozens of URLs, depth is forgiving — authority enters in many places, so a deep page is often one or two hops from somewhere with its own external links. Where one URL holds the entire external profile, every internal path starts from the same place. Depth is no longer one factor among many; it is the whole distribution graph.

Most advice states this backwards, treating internal linking as something you do after you have authority in order to allocate it. For small sites the reverse is closer to true: internal linking is what makes the little authority you have do any work at all. Crawl behaviour reinforces it, because finite crawl attention follows link structure — hard-to-reach pages get crawled less, so your edits to them get noticed later, on exactly the pages that need editing.

What hub-and-spoke actually looks like

Hub-and-spoke is a site structure in which one page per topic serves as the canonical overview, every detailed page on that topic links up to it, and the hub links down to all of them. Described that way it sounds like a table of contents, and a lot of implementations are exactly that — which is why they fail.

The hub has to earn its place

A hub page that exists only to hold links is a doorway page with better manners. It will not attract links, rank, or get cited. The hub has to be the page you would send someone to if they asked you to explain the topic from scratch: the definitions, the decision framework, the comparison of approaches, the honest statement of when each applies. Links to the spokes then follow naturally, because a thorough overview keeps reaching points where the detail belongs elsewhere.

Practical test: delete every link out of your hub page. Is what remains still worth reading? If not, you built an index, not a hub.

Spokes link up, and sideways

Every spoke does two things. It links up to the hub, usually early, in a sentence that explains why someone would go there. And it links sideways to two or three genuine siblings — pages a reader of this page would plausibly want next, not pages that share a keyword. Two or three is deliberate: a spoke linking to twelve siblings is a menu, and menus get ignored. The constraint forces you to decide what the actual next step is, and that decision is the value.

The hub gets the internal links you can spare

Because the hub is where authority pools before it distributes, the hub is the page that deserves links from your homepage, your navigation, and your service pages. Link to hubs from high-level pages; link to spokes from hubs and from each other. That is the entire architecture in one sentence.

How to write anchor text that does work

Anchor text is the visible, clickable text of a link, and it is the strongest single signal you give about what the destination page contains. Readers and search engines use it the same way: as a prediction of what is on the other side. Which makes the failure modes obvious once stated:

• "Click here", "read more", "this post". These describe the act of clicking, not the destination. Wasted signal, and worse for screen reader users, who often navigate by pulling up a list of links with no surrounding sentence.
• The exact same phrase in every link. Forty pages linking to your hub with an identical anchor is a legible pattern, and not the one natural editorial linking produces. Vary it the way you would in normal writing.
• Anchor text that oversells. "The definitive guide to X" pointing at a 600-word post trains readers not to trust your links, which costs more than the keyword gains.

What works is unglamorous: write the sentence you would write anyway, then link the words in it that name the destination. If the destination is a post about comparison pages, the anchor is something like comparison pages that get cited by AI assistants — descriptive, varied, part of the sentence rather than bolted onto it. The test is whether the link still makes sense with the surrounding text removed, because that is how automated systems often encounter it.

Linking patterns compared

Most sites run one of four structures, usually without having chosen it. They differ sharply on the three things that matter: how deep pages end up, how well equity distributes, and whether a human gets anything out of it.

Tag sprawl deserves a specific warning because it is the one people adopt as a fix. Adding thirty tags feels like adding structure, and on paper it does flatten the graph. But it flattens it through pages with no reason to exist, each holding two or three posts, many overlapping heavily. You have not concentrated anything; you have spread the same authority over forty more URLs and handed readers a set of dead ends.

The mistakes that waste internal links

Four habits account for most of the wasted internal linking we see, and all four feel productive while you are doing them.

Footer link dumps

Twenty links to popular posts in the footer, on every page. Site-wide links are the weakest kind of internal link precisely because they are site-wide: the same twenty appear under a post about pricing and a post about hiring, so they carry no contextual signal. They also inflate the outbound link count of every page, diluting the contextual links you care about. Keep the footer for navigation people need.

Linking every mention of a term

If "internal linking" appears nine times in a post and all nine link to the same page, eight of them do nothing except make the page harder to read. Link the first substantive mention, in the sentence where the reader would plausibly want to leave. Once.

Tag pages that duplicate each other

If two tag pages list substantially the same posts, at least one is costing you and helping nobody. Merge them, or remove them from the index.

Links to thin pages

Every link is an assertion that the destination is worth a reader's time. Pointing at a 300-word post that answers nothing spends your credibility and your crawl attention on a page that cannot repay either. If a page is not worth linking to, that is useful information about the page.

How to retrofit an archive you already have

Nobody reading this is starting from zero. You have an archive with the flat structure described at the top, and retrofitting a hundred posts by hand is real work. Sequence it so it pays as you go.

Two notes on execution. Work in batches you can finish — ten posts at a time, not "the archive". And re-run the crawl after each batch, because watching the inbound-link count on your hubs climb is the only feedback you will get for weeks, and it is a genuine leading indicator.

Why this compounds with AI retrieval

Everything above is standard search reasoning. There is a second reason internal linking has become more important rather than less, and it is about how AI assistants select what to cite.

Retrieval systems pull passages, not whole pages, and they have to establish that a passage is credible and on-topic before using it. The available signals are the passage, the page it sits on, and the site context around that page. A page nothing links to has no site context. It is a floating document with no demonstrated relationship to anything else you have published, and no evidence its own author considered it part of a body of work. A page in a cluster is different: it links up to an overview, sideways to related treatments, and is linked back from both. That structure is a legible claim that this site covers this topic deliberately — the same claim that makes entity-level clarity about who you are work, applied at topic level instead of brand level.

It also affects what an assistant can say about you, because related pages in the same cluster are often what it draws on to characterise a source. Orphans get cited in isolation or not at all. Clusters get cited as part of a coherent body of work. The wider implications for structuring content for these systems are in our guide to optimising for Google's AI Mode.

Common questions

How many internal links should a blog post have? There is no correct number, but a useful working range for a 2,000-word post is three to eight contextual links in the body, plus one up-link to the topic hub. The constraint that matters is relevance, not count: every link should point somewhere a reader of that specific paragraph might genuinely want to go. If you cannot justify a link that way, it is filler.

Do internal links help if my site has no backlinks at all? Yes, though differently. With no external authority to distribute, internal links still control crawl discovery, click depth and topical structure — all of which affect whether your pages get indexed and understood at all. They also make the site usable, which affects engagement. What they cannot do is manufacture authority that does not exist; they determine how whatever you have gets used.

What counts as an orphan page? Strictly, a page with zero internal links pointing to it, reachable only via the sitemap or a direct URL. Practically, treat any page whose only internal link comes from a deep paginated archive as an orphan too, because it behaves like one: minimal crawl frequency, negligible equity, and no topical context. Both categories show up in the same crawl report.

Should I remove my tag pages? Not automatically. Keep tags that correspond to a topic someone would actually search for and that list enough posts to be useful; those are effectively hubs already and are worth upgrading into real ones. Remove or de-index tags with two or three posts, or tags that list substantially the same posts as another tag. The test is whether a stranger landing on that page gets anything from it.

Does anchor text still matter, or is it outdated advice? It matters, and it matters for internal links specifically because you control all of them. Anchor text tells both readers and automated systems what to expect at the destination. What is outdated is optimising it aggressively — identical keyword-stuffed anchors across an entire site are a pattern, not a signal. Descriptive and varied is the whole instruction.

How long before internal linking changes show results? Longer than you would like, and the honest answer is that it depends on how often your site gets crawled — which is itself partly determined by the structure you are fixing. Expect re-crawling of the changed pages first, then gradual movement. Treat the internal inbound-link counts and click depths in your own crawl as the near-term measure, since those change the day you ship.