Google Is Not Reading Your Site Evenly
There’s a common mental model where Google crawls a site, reads everything, and weights each page individually on its own merits. That’s not wrong exactly — it just stops at the wrong level of abstraction. What Google is actually building is a site-level model: a representation of what your site is fundamentally about, at what depth, for which audience. Individual page signals feed that model, but the model itself shapes how individual pages rank.
Inside that model, not all pages are weighted equally as evidence of topical expertise. Some pages get treated as anchors — the core of what your site claims to know. Others get read as supporting material, peripheral content, or noise. Which pages fall into which bucket is not purely a function of how much effort you put into them.
What “Topical Core” Actually Means Here
This isn’t a formal Google term. But the underlying behavior is real and consistent enough to be worth naming. Google’s quality systems appear to identify a subset of pages on any given site that carry disproportionate weight in establishing topical authority. These pages share a few characteristics:
- They attract the highest concentration of external links that are topically relevant — not just high-authority generically.
- They receive internal link equity from multiple sections of the site, not just from a single hub page.
- They match query types where the site has demonstrated historical click satisfaction: users land, don’t immediately bounce, and don’t return to requery.
- They’ve accumulated entity associations in Google’s knowledge graph that connect the page to the topic cluster the site claims to own.
The problem: most content-marketing sites have accidentally assigned these signals to the wrong pages.
The Canonical Failure Mode
Here’s the pattern. A site publishes a high-volume keyword post early — something like “what is [core topic]” — and that page earns links because it was useful and first. Over time, the team publishes deeper, more genuinely expert content on the same topic. But the old introductory post keeps compounding its link equity and internal link count, because it’s still the most-linked page from other sites, and internal navigation still points to it prominently.
Result: Google’s topical core model for that site anchors around a shallow introductory page. The deeper, more authoritative content gets read as supplementary. The site ranks for awareness-level queries but struggles to break into competitive, high-intent queries where the deeper content is actually more relevant.
This is a different failure than keyword cannibalization. The pages aren’t competing for the same query. The wrong page has become the topical anchor, and the signals that would establish depth are concentrated in secondary content Google treats as peripheral.
How to Identify Which Pages Google Has Chosen as Your Core
You can’t ask Google directly. You can triangulate from three sources:
1. GSC Impressions vs. Internal Link Weight
Pull your top pages by impressions in Google Search Console, filtered to non-branded queries. Then cross-reference against your internal link graph — specifically, which pages receive the most unique linking pages internally (not just the most raw links). Where these two lists diverge dramatically, you have a signal: the pages Google is treating as core may not be the pages you’ve deliberately elevated through internal linking.
2. External Link Topical Relevance, Not Just Volume
Standard link audits count links or measure authority. What you actually want is the topical relevance of the linking page to your target topic cluster. A link from a generically high-authority page in an unrelated niche does much less for topical core signal than a link from a mid-authority page in your exact niche. Tools like Ahrefs give you referring page categories — that’s the dimension worth sorting by, not DR.
3. Entity Association via Knowledge Graph Cross-Reference
Search your core topic entities in Google and look at what surfaces in related entities, PAA boxes, and knowledge panels. Then check which of your pages rank for queries that include those related entities. The pages appearing there are getting entity association credit. If those aren’t the pages you’d call your best content, that’s diagnostic.
Why Information Architecture Is the Root Cause
Most sites structure their IA for users — logical categories, breadcrumbs, faceted navigation. Correct for UX. But Google’s topical model is built more from link signal patterns than from navigation labels. A category page labeled “Advanced [Topic]” does not communicate depth to Google if the internal links from that category point to posts that are two years old and never updated. The label is semantically inert. The link pattern is not.
This is why siloing alone doesn’t fix topical authority problems. You can create perfectly segmented content clusters and still have your topical core defined by whichever post happened to earn links years ago.
Fixing It Without Burning the Site Down
The instinct is usually to create a new pillar page and redirect the old one, or consolidate. Sometimes that’s right. Often the better move is more surgical:
Step 1: Deliberately reroute internal link equity. Identify the page you want Google to treat as your topical anchor. Audit every page on the site that links to the current unintentional anchor and add a link to the target page from those same pages — ideally higher in the content, with more specific anchor text. You don’t have to remove the existing links. You’re adding competing signals that point to the right page.
Step 2: Upgrade the target page’s entity density. Not keyword stuffing. Entity density means: does this page reference the full set of concepts, entities, and sub-topics that Google associates with your target topic? Use the PAA box and related searches for your core queries to map the entity set. Then audit whether your target page covers those entities substantively, or just mentions them in passing.
Step 3: Earn one genuinely topically-relevant external link to the target page. Not a high-DA link in the abstract. A link from a page that is itself topically associated with your cluster. One well-placed topically-relevant external link does more for shifting topical core signal than a dozen generic authority links.
Step 4: Track GSC impressions on the target page over 60–90 days, not rankings. Topical core shifts don’t show up as immediate ranking jumps. They show up as broader impression coverage across the topic cluster, including queries the page wasn’t previously appearing for.
A Caveat Worth Naming
This is not a one-time fix with clean feedback. Google’s quality systems update on their own schedule, and topical model shifts often don’t correlate neatly to any single action you took. If you reroute internal links and upgrade a page’s entity coverage and see no movement in 90 days, that doesn’t necessarily mean the model didn’t shift — the site may have other constraints binding first: authority floor, thin supporting content, index coverage gaps. Don’t interpret silence as failure too quickly.
Also worth noting: sites with a small number of pages in a topic cluster tend to have less distinct topical cores in Google’s model, because there isn’t enough signal differentiation to work from. For smaller sites, the intervention is usually to build more genuinely deep supporting content before trying to optimize which page anchors the cluster.
The Practical Question to Ask Right Now
Pull your GSC data, filter to your most important topic cluster, and find the page getting the most impressions for that cluster. Is that actually the page you’d hand to a skeptical expert to prove your site knows this topic deeply? If the high-impression page is an overview post and your real expertise is sitting three clicks deeper with half the internal links, that gap is your topical core problem. Fix the anchor first. Everything else is downstream of it.