A new Search Engine Land case study argues that publishers can improve organic visibility by explicitly connecting their content archives to entities in Google’s Knowledge Graph and reorganizing internal links around those entities.

Key Facts

• Claims entity SEO increased organic search traffic by +23.1%
• Search Engine Land installed TopicalBoost, an entity SEO system, on March 2, 2026.
• The implementation used natural language processing to identify entities within articles.
• Those entities were connected to Google Knowledge Graph identifiers through mainEntity, about and mentions schema properties.
• TopicalBoost also created internal links from articles to entity-focused topic archive pages.
• Entity schema was initially added to new articles, expanded to more than 30,000 archived URLs on March 23 and followed by frontend topic links on March 30.
• By the end of May, Ahrefs estimated Search Engine Land’s organic traffic at 113% of its pre-installation baseline, while the editorial comparison sites averaged approximately 70%.
• The author calculated an approximately 43-percentage-point performance gap between Search Engine Land and its editorial peers.
• Search Engine Land’s measured AI Overview citation rate increased from 1.45% to 2.98%, although other SEO websites also experienced substantial citation growth during the period.
• The evidence makes a meaningful effect plausible, but the study does not conclusively prove that entity schema caused the gains.

What Search Engine Land Implemented

Blomquist describes TopicalBoost as having two primary components: a “Disambiguation Layer” and an “Authority Routing Layer.” That terminology sounds more novel than the underlying techniques, but it is a useful way to separate the implementation.

1. Entity identification and disambiguation

TopicalBoost scans an article and identifies people, organizations, products, places and concepts. Instead of recording “Google” or “Knowledge Graph” as ordinary text strings, it associates them with specific machine-readable entities.

In the Search Engine Land source code, those entities appear inside the JSON-LD graph produced by Yoast SEO. The article’s WebPage node contains extensive about and mentions arrays. Individual entities include properties such as:

• A Knowledge Graph-style identifier in the @id field.
• The entity’s canonical name and description.
• Wikidata references.
• Official websites and social profiles.
• Additional structured properties such as founding dates, locations or entity types.

This appears to be an extension of Yoast’s existing schema graph rather than a competing block of disconnected structured data. TopicalBoost’s own website confirms that it is designed to work alongside Yoast or Rank Math rather than replace them.

The source also loads assets from a WordPress plugin identified as ttd-topics. That strongly suggests TopicalBoost is operating through a custom or branded WordPress plugin that handles entity assignments and topic presentation while modifying or filtering Yoast’s schema output.

What the public source does not prove is exactly how the entity matching occurs. The presence of Knowledge Graph identifiers makes it likely that TopicalBoost uses a Google entity service, Knowledge Graph lookup data or another entity database mapped to Google IDs. It may use Google’s Knowledge Graph Search API, but that cannot be confirmed merely by viewing the rendered page.

2. Internal links to topic pages

The second layer is arguably more important.

TopicalBoost creates or enhances pages dedicated to specific entities and then links relevant articles to those pages. An article mentioning YouTube, for example, can contribute an internal link to a YouTube topic page that collects Search Engine Land’s coverage of that entity.

This is different from simply attaching a broad “Technology” category to an article. The destination pages can represent specific people, companies, products and concepts that correspond more closely with actual search demand.

The resulting structure has several potential benefits:

• It gives crawlers additional paths into older articles.
• It reduces the number of archive pages that are weakly connected or effectively orphaned.
• It concentrates internal PageRank around subjects the publication covers repeatedly.
• It creates clearer relationships between articles and their primary or secondary topics.
• It produces useful landing pages for entity-based and long-tail searches.

This is important because schema alone is unlikely to explain a major traffic increase. Google is already capable of identifying many prominent entities in ordinary text. The combination of entity disambiguation, archive restructuring, improved crawl paths and thousands of new internal links is a much more credible mechanism.

What the Case Study Claims

Search Engine Land compared its estimated organic performance with seven other SEO websites over 13 weeks following installation.

According to the article, Ahrefs estimated that Search Engine Land finished May at 113% of its pre-installation traffic baseline. The two editorial peers in the comparison finished at an average of approximately 70%. Semrush reportedly showed a similar overall direction and magnitude.

The study also attempted to eliminate several obvious alternative explanations. Referring-domain counts remained within approximately 3%, domain ratings were flat and a later navigation change occurred after the initial increase had begun.

The most interesting evidence was not the sitewide traffic chart, but the distribution of ranking changes. The strongest gains reportedly appeared among core SEO topics with deep coverage. Off-topic segments generally performed worse. That is reasonably consistent with a system designed to consolidate signals around frequently covered entities.

Search Engine Land’s traffic count also grew faster than its estimated traffic value: 13% compared with 6%. Blomquist interprets that as evidence that the gains came primarily from informational and long-tail searches rather than expensive commercial keywords. Again, that fits the proposed mechanism.

Where Skepticism Is Still Warranted

The case study is persuasive enough to justify further testing, but it is not a controlled experiment.

First, the author is evaluating a product he co-created. That does not invalidate the findings, but independent replication would carry more weight.

Second, Ahrefs and Semrush traffic figures are estimates rather than Search Engine Land’s first-party Google Search Console data. Agreement between two tools is helpful, but both are modeling search visibility rather than measuring actual organic sessions or clicks.

Third, March 2026 included Google updates and broad changes in search behavior. A comparison group reduces that problem but cannot remove it. Core updates frequently affect ostensibly similar websites differently.

Fourth, the rollout bundled multiple changes together. The site received entity schema, revised topic associations, new archive pages, widespread internal links and potentially improved crawlability. The study cannot tell us which component produced what percentage of the result.

The AI Overview evidence is particularly difficult to isolate. Search Engine Land’s citation rate more than doubled, but Search Engine Journal also increased substantially and the SaaS-backed SEO sites grew even faster as a group. The implementation may have helped, but the data does not prove that Knowledge Graph schema directly caused Google to cite Search Engine Land more often.

The article’s strongest conclusion is therefore not that schema markup created a 43-point advantage. It is that a large-scale reorganization of entities, archive pages and internal links plausibly improved how efficiently Google discovered, grouped and evaluated Search Engine Land’s existing coverage.

How TopicalBoost Compares With InLinks

InLinks is the closest established comparison. Both systems use entity extraction rather than relying entirely on keyword matching, and both can automate internal links and deploy entity-oriented structured data.

The philosophical difference is in the graph being used and the architecture being emphasized.

InLinks has developed its own knowledge graph and semantic analyzer. It identifies entities, maps relationships between them, recommends content improvements and creates contextual links between relevant pages. It is a broader semantic SEO platform with content briefs, topic analysis and planning features.

TopicalBoost is more tightly focused on publishers. Its pitch is that articles should be explicitly connected to entities recognized by Google, while accumulated authority should be routed through dedicated topic pages. It appears less concerned with producing an all-purpose content optimization score and more concerned with transforming a large archive into a structured, entity-based publishing system.

There is substantial overlap, but TopicalBoost’s topic-page architecture and editorial workflow appear more central to the product than they are in many general-purpose entity SEO tools.

How It Compares With BrightEdge Autopilot

BrightEdge still markets the product under the Autopilot name.

Autopilot can analyze content and search opportunities, build internal-link clusters and continuously adjust optimizations through integrations with enterprise content management systems. In that respect, its internal-linking function is similar to TopicalBoost’s Authority Routing Layer.

The primary differences are market and scope. BrightEdge is an enterprise SEO platform drawing on its own keyword, ranking and competitive datasets. Autopilot can address internal linking, metadata, technical errors and other optimization opportunities. Its decisions are generally driven by search-performance and opportunity data.

TopicalBoost is narrower and more entity-centric. It organizes a publication around Knowledge Graph concepts and entity topic pages. BrightEdge may decide that a page needs links because it presents an identifiable ranking opportunity. TopicalBoost is more likely to create a link because two pieces of content belong to the same identifiable entity cluster.

For large ecommerce sites, BrightEdge’s performance-driven approach may be more naturally aligned with category and revenue priorities. For a news or research publisher with 50,000 articles about thousands of people and organizations, TopicalBoost’s model may be the cleaner fit.

How It Differs From Related-Content Plugins

Ordinary WordPress related-post plugins usually compare categories, tags, titles or term frequency and then display several suggested articles beneath a post. They can improve discovery and create additional internal links, but most do not establish a durable entity model.

A typical related-content plugin answers, “Which other articles look similar to this one?”

An entity system attempts to answer:

• What real-world subject is this article primarily about?
• Which secondary entities does it discuss?
• Which page represents each entity on this website?
• Where should internal authority be consolidated?
• How should those relationships be expressed to machines?

Related-content tools can be useful, but they often produce article-to-article links that shift as algorithms or inventories change. TopicalBoost creates a more stable article-to-entity-to-article architecture.

How Publishers Can Build a Similar System

A publisher does not necessarily need a commercial platform to reproduce the basic strategy, although building it reliably at scale is not trivial.

1. Create an entity database. Store a canonical name, aliases, entity type, Knowledge Graph or Wikidata identifier, description, official URL and relevant sameAs references.
2. Extract entities from content. Use an NLP service or named-entity-recognition model to identify people, organizations, locations, products and concepts in each article.
3. Add editorial review. Editors should be able to approve entities and distinguish the primary subject from closely related and incidental mentions. Fully automated extraction will produce errors.
4. Create one canonical page per useful entity. Avoid generating thin pages for every name ever mentioned. A topic page should contain unique introductory copy, relevant articles, pagination, proper canonicals and enough coverage to deserve indexation.
5. Add contextual or visible topic links. Link articles to their canonical topic pages. A restrained “Topics in this article” module is often safer and easier to maintain than inserting excessive links into article prose.
6. Extend the existing schema graph. Add approved entities through mainEntity, about and mentions. Reuse consistent identifiers across every article and topic page.
7. Connect topic pages back to their articles. Use crawlable HTML links rather than relying on client-side widgets that search engines may not consistently process.
8. Control index quality. Noindex, merge or suppress entities with insufficient coverage. Otherwise, the system can create thousands of low-value archive pages.
9. Measure the rollout in phases. Record Search Console clicks, rankings, crawl activity, indexation, topic-page traffic and AI citation visibility before making changes. Staged releases make causal analysis easier.

For WordPress, this could be built with a custom taxonomy, term metadata, an entity-recognition API, custom topic templates and filters that extend the Yoast schema graph. The hard part is not printing JSON-LD. It is accurately resolving ambiguous names, preventing duplicate entities and deciding which topic pages deserve internal links and indexation.

Should Publishers Implement It?

The article was written by Cord Blomquist, co-creator of TopicalBoost, the platform used in the test. Blomquist deserves credit for presenting considerably more data than the typical SEO case study. However, because the study’s author also created the product being evaluated, its causal claims deserve careful scrutiny.

The case that Search Engine Land benefited from the implementation is plausible. The timing fits, the affected keyword groups fit and the combination of better archive discovery and concentrated internal linking has a sound technical basis.

The article goes too far when it treats TopicalBoost as the proven direct cause of the entire performance gap. The available evidence supports “likely contributed materially” more comfortably than “direct consequence.”

Even with that qualification, large publishers should take this strategy seriously. A publication with tens of thousands of articles often has substantial authority trapped inside old URLs, inconsistent tags and broad category archives. Converting that archive into curated entity pages with crawlable internal links is a rational response to both traditional search and AI-assisted retrieval.

Ecommerce websites can also benefit, especially those with large editorial sections or complicated catalogs. Their entities may be brands, product families, materials, technologies, ingredients, compatibility standards or use cases. However, ecommerce implementations should generally route authority toward useful category, brand and buying-guide pages rather than automatically creating indexable archives for every entity detected in product descriptions.

Service businesses and smaller sites do not need to manufacture thousands of topic pages. They are usually better served by a well-designed taxonomy, strong hub pages, thoughtful internal linking and accurate organization, service and author schema.

The broader lesson is not that adding Knowledge Graph IDs will magically restore lost traffic. It is that large websites should stop treating archives as chronological storage. Content should be organized into a coherent, machine-readable system of subjects, relationships and internal links.

That principle predates TopicalBoost, but the Search Engine Land implementation provides an interesting and reasonably credible demonstration of what can happen when it is applied across an entire publication.

GET A FREE SEO REVIEW