Topical Authority: How Dilution Breaks Rankings and How to Fix It

Topical authority is the share of a site’s content, internal links and search engagement that concentrates on a single subject area. It decides which sites Google treats as a credible source on that subject, and it is not a number in a tool. It can be measured, it can be damaged by publishing outside the core subject, and it can be rebuilt in a defined sequence: prune, consolidate, then publish depth.

The damage usually arrives quietly. A site publishes steadily for years, drifting across subjects that looked commercially adjacent at the time, and nothing appears to break. Then a core update lands and visibility falls across pages that were never touched, including pages that had ranked well for years. The pages did not change. What changed was the weight Google applies to the domain as a whole when deciding whether this site is a credible answer on any given subject.

What Topical Authority Actually Measures

Topical authority measures concentration, not volume. A site with four hundred pages spread across nine unrelated subjects carries less subject-level credibility than a site with eighty pages that all serve one commercial core. Google’s ranking systems assess the site as a whole when judging whether a page deserves to rank, which means the pages around a page affect how that page performs.

Three inputs drive it, and all three have to point the same way.

Content Concentration

The first input is the proportion of indexed URLs that sit inside the core subject. This is the one most sites get wrong, because content decisions are made one article at a time. A guide on a tangential subject looks harmless on its own. Fifty of them, published across four years by three different agencies and two in-house marketers, become a structural problem that no single article caused. Anyone reviewing the output of a long-running content marketing programme should be reading the library as a portfolio, not as a series of individual commissions.

Internal Link Concentration

The second input is where internal links point. Internal linking tells Google which pages a site considers important and which subjects it organises itself around. When links are distributed evenly across every subject a site has ever covered, that signal flattens. When they concentrate on a commercial core and its supporting articles, the core reads as the site’s actual specialism. Off-topic content does not simply sit inert; it absorbs internal links, crawl attention and authority that the core needed.

Engagement Concentration

The third input is which subjects earn genuine search engagement. A site can hold hundreds of pages on a subject and still not be treated as authoritative on it if none of those pages earn clicks. Impressions without clicks are the clearest early signal of dilution: Google is willing to test the page in results and users consistently pick someone else. Handled properly, an SEO services engagement treats that pattern as diagnostic rather than as a reason to publish more.

Topical Authority vs Domain Authority

These are separate ideas that get conflated constantly, and the confusion is expensive because it points investment in the wrong direction.

Google has said repeatedly that domain authority scores are not ranking inputs. The reason those scores still correlate with rankings is that the things they proxy, link quality and site-wide trust, do matter, through Google’s own systems rather than any vendor’s number. A site can hold a high domain authority score and still lose rankings in a core update if its subject concentration has drifted, which is precisely the failure mode most owners find inexplicable.

How Dilution Damages Rankings

Dilution damages rankings through reweighting, not penalties. There is no message in Search Console. Nothing is manually actioned. The site is simply reassessed against a slightly different set of weights, and a domain that reads as a generalist publisher loses to domains that read as specialists on each individual subject.

Three mechanisms do the work.

Classification drift. Search engines have to decide what a site is about before deciding whether it deserves to answer a query. A library split across nine subjects makes that decision ambiguous. Ambiguity resolves against the site, and it resolves across every subject at once, including the ones that were performing.

Crawl and link dilution. Every off-topic URL consumes crawl allocation and holds internal links that could have supported the commercial core. On a large site this compounds quietly: the pages that most need internal support are competing for it with pages that will never earn a click.

Quality averaging. Off-topic content tends to be the weakest content on a site, because it was written outside the team’s actual expertise. Google’s published guidance on creating helpful, people-first content is explicit that the assessment applies at site level as well as page level. Thin content on subjects a business has no standing in pulls the site-level assessment down and takes the good pages with it.

The pattern is consistent enough to be predictable. Sites that lose visibility in core updates without any obvious page-level cause are usually sites whose subject concentration fell below the point where they read as a specialist on anything.

How to Audit Topical Dilution

The audit is mechanical. It takes patience rather than sophistication, and on a library of several thousand URLs it takes a structured spreadsheet and a few days of disciplined classification. Anyone running this at scale should work from a documented content audit framework rather than improvising the method each time, because consistency of classification matters more than the cleverness of the categories.

Step 1: Map Every URL to One Topic

Export every indexed URL and assign each one a single topic. One topic per URL, no exceptions, no hybrids. The discipline of forcing a single assignment is the point: a URL that resists classification is usually a URL that resists ranking, because search engines faced the same ambiguity.

Then sort every topic into three buckets.

The uncomfortable part of this exercise is that “adjacent” quietly absorbs everything a team does not want to delete. Apply one test: would a buyer of the core service plausibly read this page on their way to an enquiry? If the honest answer is no, the page is off-topic regardless of how well it was written.

Step 2: Measure the Core-to-Off-Topic Ratio

Count URLs by bucket and calculate what proportion of the indexed library sits in the core. This single ratio is the most useful diagnostic in the whole exercise, and it is almost always worse than the team expects. Run the same count weighted by clicks and by impressions. A site where the core holds a minority of URLs but the large majority of clicks has a pruning problem with a clear solution. A site where the core holds neither is in a harder position, and the rebuild will take longer.

Step 3: Check Which Topics Actually Earn the Clicks

Pull clicks and impressions by URL from Search Console over a twelve-month window and aggregate by topic. Two patterns matter more than the totals.

The first is high impressions with negligible clicks. Google is surfacing these pages deep in results and users are choosing other sources. At the topic level, a subject with substantial impressions and almost no clicks is a subject where the site is being tested and consistently rejected. That is dilution showing up in data rather than in theory.

The second is clicks concentrated in a handful of URLs that have nothing to do with the commercial core. This happens more than most owners realise, and it is a trap. Those pages feel valuable because they carry traffic, but traffic that never becomes an enquiry is a cost centre with good analytics.

Step 4: Cross-Reference AI Citations

Bing Webmaster Tools reports which pages are cited in AI answers, and that data changes content pruning decisions. A page with weak Google performance can still be a heavily cited AI asset, usually because its tables, ranked lists or numbered steps are easy to extract. Check citation counts before anything is removed. Structural elements driving citations should survive the rebuild intact even where the surrounding copy is rewritten.

The Rebuild Sequence: Prune, Consolidate, Publish Depth

Sequence matters here more than effort. Teams that publish depth first, while the diluting content stays live, spend heavily and move slowly, because the new work is entering a library that still reads as unfocused.

Prune First

Removing content feels like destroying assets, and it is the step that gets deferred longest. Off-topic URLs with no clicks, no citations and no links are removed or redirected to the closest relevant core page. Pages with genuine external links are redirected rather than deleted, so the link equity moves inward instead of evaporating. Pruning is not a mass deletion exercise; it is a URL-by-URL judgement applied consistently, and on a large library it runs in batches over weeks with position monitoring between each pass.

“The hardest conversation in any content recovery is telling a client that pages they paid for are actively holding the site back,” says Ciaran Connolly, founder of ProfileTree. “Nobody wants to hear that publishing more was the wrong answer. Once the off-topic material is cleared and the internal links point at the commercial core, the same articles that sat at position sixty start behaving like they were written by a specialist, because now the site around them reads like one.”

Consolidate Second

Once the off-topic material is gone, overlapping core pages come next. Sites that have published on one subject for years accumulate near-duplicates: three articles answering the same question from slightly different angles, each ranking weakly, none ranking well. Pick the strongest URL as the survivor, fold the genuinely useful material from the others into it, and 301 the losers to it. Merging is where most of the near-term ranking recovery happens, because it converts several weak signals into one strong page without producing a single new word.

Publish Depth Last

New publishing starts only after the library is clean and consolidated. Depth means covering the core subject and its subtopics thoroughly enough that a reader has no reason to look elsewhere, with each new page linking to the pillar and to the siblings that expand on it. Publishing into a concentrated library performs differently to publishing into a diluted one, which is why this step comes third rather than first.

What Recovery Looks Like and How Long It Takes

Recovery is uneven and it is slower than the damage was. Expect the pattern rather than a date.

Consolidation moves first, usually within weeks, because merged pages resolve competing signals that Google can re-evaluate on the next crawl. Pruning shows up next as crawl allocation shifts toward the pages that remain, visible in Search Console crawl stats before it is visible in positions. Site-level reassessment is the slow part. Sites reweighted downward in a core update frequently do not recover fully until a later update reassesses the domain, which can mean several months of doing the right things with limited visible reward.

That gap is where most recoveries are abandoned. Teams prune, consolidate, see partial movement, lose confidence and start publishing across subjects again to chase traffic, which resets the concentration they just spent months building. Holding the line through the flat period is the actual difficulty of this work, and it is more a management problem than a technical one.

Final Thoughts

Topical authority is a resource allocation decision disguised as an SEO concept. Every page a site publishes either concentrates the subject signal or dilutes it, and there is no neutral option. The sites that survive core updates are not the sites with the most content or the strongest link profiles; they are the sites whose libraries make it obvious what they are for.

For most SMEs the practical implication is narrower than it sounds. Pick the subject the business actually sells, audit honestly against it, remove what does not serve it, merge what duplicates it, and then go deep. That sequence is unglamorous and it works, and the discipline it requires is mostly the discipline of not publishing the article that felt like a good idea in a meeting.

Frequently Asked Questions