Introduction
Internal authority is often described as something you can redistribute at will. Add links, tweak navigation, and the weight should flow.
In practice, authority inside a site behaves more like routing priority than a transferable asset. It moves along paths that are stable, repeatedly traversed, and cheap for the system to interpret. Everything else leaks.
I’m writing this because most internal linking debates stall at the wrong layer. They focus on counts and placements, while the real constraints live in traversal frequency, consolidation cost, and intent stability.
Authority is constrained by interpretability
Authority does not move evenly. It moves where the site is readable.
If every step in the internal graph forces the system to resolve ambiguity — mixed intents, competing canonicals, category pages that behave like SERPs — consolidation slows down. Signals become fragile. Links exist, but they don’t reinforce.
Google engineers have hinted at this for years. John Mueller has repeatedly said that internal links help search engines understand importance, not because of volume, but because of context and repetition. Gary Illyes has framed it even more bluntly: systems “focus on what they see often and understand well.” That’s not a ranking trick. That’s scheduling.
This is where taxonomy & intent structure stops being an IA discussion and becomes an authority discussion. Clear boundaries reduce decision cost. Reduced decision cost makes signals stick.
The graph you design vs the graph that exists
Most teams imagine a clean hierarchy: homepage → categories → subcategories → content.
The crawler does not see that diagram. It sees URL states and how often they are re-encountered.
On sites above ~100k indexable URLs, crawl data typically shows a steep drop-off after the first few layers. Pages deeper than level 4–5 often receive 3–10× fewer revisits per month than top-level hubs. Depth itself is not fatal, but depth combined with low traversal frequency is.
This is the operational side of URL depth vs crawl frequency. Authority cannot accumulate where revisit probability is low.
Where authority actually accumulates
On large sites, authority concentrates quickly.
Across ecommerce, media, and SaaS documentation sites with 100k–1M URLs, it is common to find that fewer than 5% of pages account for the majority of high-frequency crawl encounters. Those pages are reprocessed often, consolidate signals quickly, and act as stable routing points.
The rest are reachable, but weakly reinforced.
| Page role | Typical revisit frequency | Authority behaviour | Common failure pattern |
|---|---|---|---|
| Hub | Daily / weekly | redistributes authority | loses focus, becomes bloated |
| Category | Weekly | sets topical boundary | turns into faceted search |
| Article | Monthly or less | consumes authority | drifts into soft-orphan state |
| Filter / parameter | Unbounded | none (should be constrained) | absorbs crawl, leaks signals |
Hubs and categories are not interchangeable
The difference between a hub and a category is mechanical, not semantic.
A hub is a routing surface. It is designed to be revisited frequently and to expose downstream nodes repeatedly. A category is a boundary. It communicates scope and limits.
When category pages are forced to act like hubs — endless pagination, filter combinations that look indexable — they stop being stable boundaries. Authority flow becomes noisy. Consolidation cost rises.
That structural distinction is the core idea behind hub vs category pages. Confusing the roles destabilises the internal graph in ways links alone cannot fix.
Authority moves through reinforcement loops
A useful model is simple:
- a URL is discovered,
- it is reinforced through repeat encounter,
- its signals consolidate,
- it becomes a reliable routing node.
Break reinforcement and nothing downstream stabilises. You can see this directly in logs: URLs linked once spike, then decay back to baseline.
This is why linking to a section once, from a low-frequency page, rarely changes anything. The path exists. The loop does not. Slight pause there — that’s the part people miss.
What gets misdiagnosed as “authority problems”
The same complaints surface again and again:
- “We added internal links and nothing changed.”
- “Only a handful of pages rank.”
- “Deep pages never gain traction.”
In most cases, links were added without creating repeatable paths. Connectivity increased. Reinforcement did not.
Conclusion
Authority inside a site does not behave like a liquid. It behaves like routing priority.
It accumulates in pages that are structurally central, frequently revisited, and intent-stable. It reaches other pages through reinforcement loops, not through one-off links. When taxonomy is ambiguous, depth correlates with low traversal frequency, and categories are forced to behave like hubs, authority flow degrades.
If you want internal authority to hold, make the site easier to interpret. Everything else is decoration.