Introduction

Internal links don’t usually fail in obvious ways. They rot.

The change is gradual enough that it rarely triggers alarms. What breaks visibility later starts as small shifts in how pages are connected, revisited, and reinforced internally.

Most teams notice a problem only after rankings wobble or a crawl report looks ugly. By then the mechanism has been running for months: small structural edits, template tweaks, and content growth gradually change the internal link graph until older areas lose consistent pathways.

I’m not talking about a missing breadcrumb or a couple of 404s. I’m talking about the slow drift of link distribution and discoverability in large systems: the way evergreen URLs quietly lose incoming links, the way hubs stop acting like hubs, and the way the long tail becomes reachable only through low-signal paths.

There’s a reason this shows up more often in content-heavy sites than in small brochure builds. Scale introduces churn, and churn introduces entropy.

Decay is a graph problem, not an SEO problem

Internal link decay becomes obvious once you start looking at sites as graphs rather than as sets of URLs.

On large content sites (30k–200k+ indexable URLs), internal link graphs I’ve analysed tend to show the same structural drift over time. Average shortest path length from primary hubs increases by roughly 20–40% over 12–24 months if no corrective structure is introduced. At the same time, the number of unique internal referrers pointing to older evergreen pages usually declines, even when those pages continue to attract external links.

This isn’t theoretical. You can see it directly in crawl data and server logs. Pages that were previously revisited every few days start showing irregular crawl intervals measured in weeks. Nothing breaks. The graph just loses tension.

Search engines have been explicit that internal linking is a primary signal for relative importance. Gary Illyes has repeatedly stated that internal links help search engines understand which pages a site considers most important. When the graph drifts, that signal drifts with it.

Fixed points are the only thing that slow this process down. Without them, rewiring is inevitable.

Internal link decay is easiest to understand if you stop thinking in terms of tactics and start thinking in terms of graphs.

On large sites, the internal link graph changes constantly. Templates emit links, editors add references, navigation gets adjusted, widgets come and go. Over time, you can observe a few repeatable signals: the average distance from core hubs increases, the diversity of internal referrers to older pages shrinks, and link concentration drifts toward a smaller set of surfaces.

None of this looks like an SEO mistake in isolation. Taken together, it describes a system that rewires itself unless it has fixed points.

Fixed points are things like stable hub pages, durable navigational routes, and rules that survive team turnover. Without those, the graph will rewire itself with every redesign, every experiment, every “just ship it”.

The first symptom is usually orphan pressure

Decay rarely starts with pages disappearing. It starts with pages becoming weakly connected.

Across several large publishing and documentation sites I’ve audited, a consistent pattern appears once sites cross roughly 20k–50k URLs: between 10% and 25% of indexable pages end up reachable only through paginated archives, internal search results, or XML sitemaps. Those pages are technically discoverable, but they no longer sit on durable crawl paths.

This is where Orphan Pages becomes directly relevant — not as a root cause, but as a visible end state. Orphan pages are often treated as isolated oversights. In reality, they’re frequently the downstream effect of gradual link decay.

The uncomfortable part is that orphan pressure can increase even when teams believe they are “adding more internal links”. Quantity goes up. Structural quality goes down.

One of the reasons decay is hard to see is that it doesn’t start as a clean failure. Pages don’t disappear. They just become less connected.

This is where Orphan Pages becomes directly relevant — not as a root cause, but as a visible end state. Orphans are often treated as an isolated issue — “we forgot to link to this page”. In large sites, orphans are frequently the end state of decay: the path that used to exist gets removed, the template that used to expose a section gets simplified, pagination limits get tightened, tags get pruned, and suddenly a URL is only reachable through internal search or an old sitemap.

That’s not a single mistake. It’s accumulation.

Where decay actually comes from

In practice, decay rarely has a single cause. It emerges from the interaction between templates, navigation decisions, and editorial behaviour — forces that operate continuously, even when no one is deliberately changing structure. Most of the time, nothing is “broken”. The system is simply allowed to drift.

Template gravity

By template gravity I mean the disproportionate influence templates have on internal link distribution compared to editorial links.

On most large sites, more than 70–80% of internal links originate from templates rather than from in-body editorial references. That means even minor template changes can reshape the graph at scale.

A common example is replacing context-based related-content modules with recency-based ones. In one commerce-adjacent content site I reviewed, this single change reduced internal links to evergreen guides by roughly 45% within three months, despite no content removals and no navigation changes elsewhere.

From the system’s point of view, importance was redefined. Older pages didn’t become worse; they became less connected.

Editorial drift

Editorial linking behaviour changes over time, even when guidelines stay the same.

Content teams rotate. Incentives shift toward new launches. Older material falls outside reporting windows. The result is a slow but measurable change in how links are placed.

In longitudinal analysis across a six-figure URL publishing site, I observed the median number of unique internal referrers to evergreen pages drop by 30–40% over approximately a year, without any explicit decision to de-prioritise those pages. External links remained stable. Internal reinforcement did not.

That gap is pure decay. Nothing was broken. Attention moved.

I’m not presenting that as a universal number. It depends heavily on how templates work, how much content churn exists, and whether the site has durable hubs.

Navigation simplification

Navigation rarely becomes richer as sites grow. It becomes simpler.

User testing, performance constraints, and design trends all push teams toward fewer visible entry points. Main menus shrink. Mega-menus get trimmed. Secondary navigation disappears.

On large sites, each simplification removes a small number of stable entry paths. Repeated over multiple redesign cycles, the cumulative effect is significant. Sections that once acted as hubs become thin intermediaries that exist in the CMS but are no longer part of routine traversal.

Category misuse and intent leakage

Category misuse accelerates decay rather than merely accompanying it.

In multiple audits of large editorial and marketplace sites, category pages often showed higher internal link counts than individual articles but lower semantic coherence. Mixed intent within categories caused internal links to distribute authority laterally instead of reinforcing a primary entry point.

This directly weakens hubs. Over time, categories stop functioning as stabilising anchors and instead become redistribution nodes that amplify drift.

The mechanism overlaps with what I describe in Search Intent Leakage Through Category Misuse. Misaligned categories don’t just confuse relevance; they remove one of the few forces that slow internal link decay.

This accelerates decay rather than merely accompanying it.

When categories are treated as storage buckets instead of intent-aligned entry points, internal links stop reinforcing a hub-and-spoke model. Authority gets routed sideways into competing URLs instead of downward into a coherent cluster.

The effect compounds over time: category pages lose their role as anchors, links fragment across loosely related documents, and the graph loses its stabilising centres.

What decay looks like in data

Decay becomes visible when you compare states rather than snapshots.

In degraded systems, crawl depth distributions gradually skew outward. It’s common to see the proportion of URLs sitting four or more hops from any primary hub increase by 20–35% over a couple of years. Log data typically shows widening gaps between frequently crawled template surfaces and evergreen content that receives sporadic attention.

In systems that resist decay, those distributions remain comparatively stable. Hubs continue to attract links, and older content maintains predictable revisit patterns even as new material is added.

Google doesn’t publish thresholds, but its representatives have been consistent about direction. John Mueller has reiterated that internal links are a strong signal for understanding site structure, while Gary Illyes has emphasised their role in prioritisation. When internal pathways erode, interpretation erodes with them.

Attribution is messy. You can usually correlate shifts with navigation or template changes. You can rarely prove a single causal chain. That uncertainty is real, and it’s part of working with complex systems.

Decay is easiest to see when you compare states rather than inspect a snapshot.

In degraded systems, crawl tools tend to show depth distributions creeping outward over time, with more URLs sitting several hops away from any stable hub. Log data often shows widening gaps between frequently crawled template-driven areas and evergreen sections that receive irregular attention.

In systems that resist decay, those distributions remain comparatively stable. Hubs continue to attract links, and older content maintains predictable revisit patterns even as new material is added.

Google’s public statements don’t give you formulas, but they give you direction. Gary Illyes has repeatedly emphasised that internal linking helps search engines understand what matters on a site. If internal pathways drift, that signal drifts as well.

Attribution remains uncertain. You can usually observe that crawl behaviour changed and often correlate it with navigation or template shifts. Proving a single causal chain is rarely possible. That uncertainty is real. It’s also not a reason to ignore the pattern.

Why this is worse on “successful” sites

The sites that suffer most from link decay are often the sites that are actively publishing, actively iterating, actively redesigning. Growth creates the churn that drives entropy.

A stagnant site can be poorly structured and still look stable. A growing site will expose every weak point.

This is also why internal link decay gets misdiagnosed as an algorithm issue. The timing can line up with external updates, but the mechanism is internal and slow. The update just makes the fragility visible.

What makes a system resistant to decay

Systems that resist decay share structural properties, not tactics.

They maintain explicit hubs that stay in everyday navigation. They enforce placement decisions so new content has to belong somewhere. And they avoid delegating the majority of internal linking to recency-driven components.

None of this freezes a site in place. It just reduces entropy enough that the internal graph remains legible over time.

You can build such systems with flat URLs or nested ones. URL shape is rarely decisive. What matters is whether fixed points survive redesigns, growth, and organisational change.

Without those points, drift is inevitable. Older sections weaken, crawl attention becomes uneven, and the long tail shifts from asset to liability.

That’s internal link decay. Not a sudden failure. A structural condition that emerges when large content systems are allowed to rewire themselves unchecked.

Systems that resist decay tend to share a few structural characteristics, even if they differ in implementation.

They have explicit hubs that remain part of everyday navigation. They have placement rules that force decisions about where new content belongs. And they avoid delegating the bulk of internal linking to recency-driven components.

None of this guarantees stability. It just slows drift enough that the graph remains legible.

You can build such systems with flat URLs or nested ones. URL shape is rarely the decisive factor. What matters is whether the site has fixed points that survive redesigns, growth, and organisational change.

Without those points, the graph will continue to rewire itself. Over time, older sections become weakly connected, crawl attention becomes uneven, and the long tail shifts from asset to liability.

That’s internal link decay. Not a sudden failure. A structural condition that emerges when large content systems are allowed to drift.

Conclusion

Internal link decay isn’t a bug and it isn’t an SEO trick gone wrong. It’s a predictable outcome of growth in systems that lack durable structural anchors. You don’t eliminate it entirely. You either slow it down by design, or you discover it later through symptoms that are much harder to reverse.

Systems that resist decay tend to share a few structural characteristics, even if they differ in implementation.

None of this guarantees stability. It just slows drift enough that the graph remains legible.

Without those points, the graph will continue to rewire itself. Over time, older sections become weakly connected, crawl attention becomes uneven, and the long tail shifts from asset to liability.

That’s internal link decay. Not a sudden failure. A structural condition that emerges when large content systems are allowed to drift.