Topical authority, declared: How Search Engine Land bucked the editorial traffic decline

SEOs seem to agree on the importance of topical authority. According to a Surfer internal survey, 88% of SEOs believe topical authority is very important to their SEO strategy. 

Google concurs. The Search Central blog has discussed how topic authority works and the 2024 Content Warehouse API leak revealed topical relevance values like siteFocusScore, siteRadius, siteEmbeddings, and pageEmbeddings. Elizabeth Tucker, Director of Product Management at Google Search, mentioned topics, topicality, and topical relevance 13 times in a 36-minute episode of Search Off the Record.

But how do we use this to improve our rankings, increase AI citations, drive traffic, and ultimately achieve our business goals? I’ve seen compelling evidence that topical authority can be improved through things like more brand mentions, content clusters, or mastering the dark art of semantic SEO. The problem I’ve found with these approaches is applying them to publishers.

A look at what topical authority is

First, let’s define a publisher as a site with thousands of pages, a regular publishing cadence, and — usually — a domain rating that would make most local, ecommerce, or SaaS SEOs incredibly jealous. You’d think that for publishers SEO would be easy, but, counterintuitively, all of those articles and backlinks can serve to muddle their topical authority. That’s because, when you write about everything and get links from everywhere, you end up being like Seinfeld, a site about nothing specific enough.

Solving this problem is a matter of unpacking the concept of topical authority. It’s made up of the two things it sounds like it would be: topics and authority. 

In its official developer API frameworks and patent filings, Google defines topics as represented by Knowledge Graph entities. And when we look at how these systems evaluate the open web, Google’s own documentation confirms that the Knowledge Graph serves as its primary database for mapping real-world facts.

Google still defines the authority, at least partially, using hyperlinks according to its How Search Works documentation, but raw link power is now checked by other mechanisms like brand mentions and how frequently original content is cited by your peers. 

Even if links were the only factor in measuring authority, for most publishers, backlinks are scattered across thousands of articles. Like an army dispersed across a battlefield, they would be a formidable force, if only they could concentrate their effects.

When viewed this way, a potential solution to publishers’ topical authority problem seems clear: connect those scattered articles to each other around topics found in the Knowledge Graph.

Connecting articles to one another and the Knowledge Graph: A case study

I was recently given the chance to do this for Search Engine Land itself. Search Engine Land installed TopicalBoost, the entity-based SEO solution I co-created with Einar Jóhannsson, on March 2, 2026. TopicalBoost uses Natural Language Processing (NLP) to extract entities from posts, tie them to the Knowledge Graph through schema markup, and route internal links between them through topic pages. You can see this in action by viewing the source code of this page and the list of linked topics below the article.

Thirteen weeks of post-install data gives us enough room to run a panel comparison against a group of seven competitors, measuring how SEL moved relative to peers exposed to the same market conditions.

Two questions structure this case study:

1. Did Search Engine Land actually outperform its peers on organic search after the install?
2. If it did, was TopicalBoost actually the cause or could it have been coincidence, a backlink surge, an algorithm tailwind, or something else?

The short answer to the first question is yes, and both Ahrefs and Semrush data agree on direction. The rest of this article will be about trying to answer the second.

Question 1: Did Search Engine Land actually outperform?

Our panel data of Search Engine Land (SEL) and seven competitors shows that most editorial SEO publishers are losing traffic. This is no surprise in the era of AI Overviews (AIO). 

Among tool sites — sites tied to a SaaS offering — the picture is more mixed: Yoast and WordStream are roughly flat, while Moz, Semrush, and Ahrefs have grown.

Daily organic traffic, Search Engine Land vs. seven peer SEO publishers

The chart above shows Ahrefs estimated daily organic traffic for the eight sites in the panel. To give a legible comparison, I indexed each site to its own baseline: the four weeks before TopicalBoost activated, January 26 through February 22, 2026. So what you’re seeing is a chart of percentage change against the baseline. The baseline is the gray area on the left. The green region on the right is after TopicalBoost’s site-wide analysis on SEL went live.

The data above from Ahrefs shows Search Engine Land’s traffic ended May at 113% of its pre-TopicalBoost baseline. The two editorial peers in the panel ranged widely, but ended May at an average of 70% of their baseline.

It’s worth noting that SEL wasn’t flat heading into March. Rather, it had been declining at roughly the same rate as Search Engine Journal (SEJ). After March 2, SEL reversed course while SEJ kept declining.

I confirmed the trend among editorial SEO sites using Semrush. The data was a bit more noisy day-to-day, so I produced a simple bar chart showing the difference between the  baseline and the end of May. The numbers are largely the same.

Semrush organic traffic change after TopicalBoost

The gap between baseline and May 31 is around 43 percentage points by both measures. Both agree on direction and on roughly the same magnitude. Editorial peers fell, but Search Engine Land didn’t.

With two sources agreeing on direction and magnitude, the question isn’t, “Is the gap real?” anymore. It’s, “What caused it?”

Question 2: How do we know it’s TopicalBoost?

To test whether or not these changes were due to TopicalBoost, we next ran five tests in two groups. First, three checks on whether the basic claim survives the obvious objections:

1. Can we rule out alternative explanations like backlinks, an algorithm change, or an unrelated site change?
2. Does the timing of SEL’s lift coincide with the TopicalBoost rollout?
3. Does the pattern of which keywords moved fit that same mechanism?

We then looked at two additional predictions the mechanism makes. Both should hold if TopicalBoost is in fact the driver:

1. Did SEL’s traffic gains come from lower-cost-per-click (CPC) informational queries as the mechanism suggests it should?
2. Did SEL’s AI Overview citation rate move with the rollout?

What we can rule out

There was no new referring domain surge that could plausibly account for the SEL-specific lift. Across the eight-site panel, referring-domain counts stayed within + or – 3% and DR was flat. 

An algorithm change is harder to rule out. Core updates don’t lift or punish a peer set uniformly — they reshuffle, sometimes rewarding one site at the expense of another in the same category. We can’t say, “No algorithm update happened, therefore TopicalBoost.” What we can say is that the lift’s timing, while somewhat confounded by March’s spam and core updates, maps to the rollout dates.

That leaves the alternative of some other change at Search Engine Land itself during the window. We identified one: The week of April 26 saw Search Engine Land change its primary navigation. Like TopicalBoost, this would also redistribute internal link equity across the site, but the timing here, too, is off. The Ahrefs traffic estimate began rising the first week of March and peaked in mid-April, which predates the nav change.

The takeaway: A backlink surge doesn’t fit. A category-wide algorithm tailwind doesn’t fit. The post-April nav change can’t account for the lift that began in early March. Still looks like our entity-SEO changes were the difference-maker.

The timing fits the mechanism

This TopicalBoost project rolled out in three phases:

• Phase 1 (March 2): JSON-LD entity schema activated for new articles only
• Phase 2 (March 23): JSON-LD entity schema applied across the entire archive of 30,000+ URLs
• Phase 3 (March 30): Frontend topic-archive page links went live

Traffic began rising in early March — coinciding with Phase 1 and reversing the previous decline — and reached 108% of baseline by March 23. From there through mid-April, it climbed to a peak of 125%. Search Engine Land’s curve in Chart 1 shows that inflection clearly: The bold blue line climbs out of the cluster after March 2 while editorial peers continue declining.

This all lines up with the phased rollout. A smaller lift arrives with the schema disambiguation layer. A larger lift arrives once the link-routing layer is in place.

The structural pattern fits the mechanism

If TopicalBoost is doing the work, keyword rankings gains shouldn’t be evenly distributed. A rising tide might lift all boats, but what TopicalBoost does is more targeted than that. We should see upward movement on topics SEL has covered deeply, less on topics SEL has covered shallowly, and no movement at all on terrain outside SEL’s editorial domain.

I built a heatmap to look for that pattern. It shows keyword position gain/loss on a grid comparing coverage depth (number of posts per topic) and topical relevance.

We measured topical relevance using a co-occurrence propagation model — think of this as PageRank for the relevance of topics to your site. The model spread relevance from seed topics — foundational SEO topics like “search engine optimization,” “Google Search,” “link building,” and “AI Overviews” — to all topics that co-occur with them frequently in SEL’s archive. Cells with fewer than 20 keywords are shown but should be treated as too sparse to interpret. Those with fewer than five have been dropped entirely.

Ranking change by topic relevance and coverage depth

Each cell shows how much more often keywords in that segment gained ranking than lost it after TopicalBoost activated. The “Core SEO topic” row shows positive movement at every coverage depth, climbing from +25pp at the 10-20 articles bucket to +65pp at the 2001+ articles bucket. The gradient runs cleanly from the shallowest topic coverage to the deepest. The “Adjacent” row is mixed (one cell shows +127pp on only 12 keywords, which is sampling noise and shouldn’t be over-read). The “Off-topic” row is mostly negative.

Now we can say that SEL — the one outlier in the editorial SEO site panel with traffic gains — rules out the obvious confounds, the timing of its traffic change fits the mechanism, and the pattern fits the mechanism.

The traffic gain came from low-CPC informational queries

Below is a traffic-value chart running the same comparison as our first chart, but this one maps traffic value (using CPC equivalency in Ahrefs). The two graphs are broadly similar, but diverge in a way that matters.

Organic traffic value, Search Engine Land vs. seven peer SEO publishers

Search Engine Land’s traffic count grew more than its traffic value (+13% vs +6% at the end of May). That gap tells us where SEL’s clicks came from: not high-CPC commercial terms, but the lower-CPC, entity-driven, informational segment of its keyword universe.

That’s exactly where TopicalBoost is designed to help. Entity disambiguation and topic-archive link routing help Google match the right page to long-tail queries — queries that are typically low-volume, low-CPC, and informational. It’s also where most of an editorial publisher’s archive sits, and where editorial peers in the panel are losing ground.

The upshot here is that SEL didn’t escape editorial decline by shifting into commercial territory. Instead, SEL gained inside the editorial segment, the same segment that’s collectively declining. That makes the lift more notable.

The same pattern could have emerged from an algorithm update. But Search Engine Journal, SEL’s closest peer in topical focus, continued declining throughout this period. A category-wide change should have lifted SEJ too.

AI Overview citation rates moved with the rollout

If TopicalBoost makes Search Engine Land’s content more recognizable at the entity and topic level, we’d expect SEL to be cited as an AI Overview source more often after the install.

The problem we ran into in testing this is that simply counting AIO citations doesn’t tell you much. It can’t tell you if a site has more AIO citations because it’s doing a special AI thing or just because it’s covering more keyword ground overall.

So instead, we divided the number of AIO citations by the number of ranking organic keywords for each site. This gives us a citation rate — how often the site makes it into AI Overviews per organic keyword. 

Using this measure, publishers can be compared regardless of site size or domain authority.

AI Overview citation rate, % of organic keywords where the domain is cited as a source in an AI Overview

Before TopicalBoost, our panel was clustered into a narrow band between 1.4% and 1.9% citation rates. Search Engine Land was at 1.45%.

After March 2, the bands separate. By May 15, Search Engine Land’s citation rate had climbed to 2.98%, more than doubling its pre-TB baseline. The editorial peer set split: Search Engine Journal also grew, ending at 2.90%, while Neil Patel held roughly flat at 1.46%. The tool sites climbed sharply: Moz to 6.11%, Semrush to 5.02%, Yoast to 4.29%, Ahrefs to 3.16%.

Cohort averages, Feb 15 → May 15:

• Editorial peers, excluding SEL: +34% (Search Engine Journal +73%, Neil Patel −5%)
• Search Engine Land: +106%
• Tool sites: +136%

Search Engine Land’s growth trajectory pulled away from the editorial group and edged toward the tool-site group’s growth. In absolute terms, SEL still sits below the tool cluster. However, its rate of change over the post-TopicalBoost window matches the SaaS tool sites far more closely than its editorial peers. This is consistent with TopicalBoost’s mechanism. Entity-level structured data and topic-archive routing make a page more legible to Google’s source-selection systems which determines the sites Google chooses to cite inside an AI Overview.

SEJ also climbed substantially over the same window, narrowing some of the gap to SEL. We can’t rule out that SEJ made structural changes of its own, or that AIO citation rose disproportionately on the search-news queries both sites compete on.

What we can say is that SEL’s growth was larger, started earlier, and tracked the TopicalBoost rollout timing. And also that Neil Patel, the one flat editorial peer, didn’t move in the same direction.

From inference to declaration

A 43-percentage-point traffic outperformance against direct editorial competitors and a 106% surge in AI Overview citations is not the result of algorithmic luck. It is the direct consequence of a publisher choosing to take control of how Google perceives its topical authority.

Recall the split we started with. Topical authority has two ingredients: topics and authority. Most publishers leave both to inference. Google has to read tens of thousands of articles and guess what the site is about. Backlinks accumulate randomly across the archive without concentrating around the topics that matter most. The bigger and older the publisher, the worse this phenomenon can get.

TopicalBoost addresses both ingredients directly:

• The Disambiguation Layer fixes the topics problem. By injecting mainEntity, about, and mentions schema, ties its content to specific IDs in Google’s Knowledge Graph. Ambiguous text strings become unique entities.
• The Authority Routing Layer fixes the authority problem. By building internal links to dedicated topic pages, SEL concentrates its scattered link equity around the entities it covers most.

Together, these two layers convert what was inferred into what is declared.

To put it another way, before March 2, Google had to scan tens of thousands of articles and infer whether SEL was an authority on Marissa Mayer. After March 2, SEL stated it directly: 396 articles about Marissa Mayer, 4,200 about YouTube, 289 about Bing Webmaster Tools, and so on. The inference problem became a lookup. Google doesn’t have to guess anymore.

The future of SEO isn’t about writing more, it’s about declaring more

And, as we’ve shown, this isn’t only about traditional organic search. Google’s own research keeps reinforcing the Knowledge Graph’s central role: the SKILL framework for infusing structured knowledge into LLMs, Talk Like a Graph on improving LLM reasoning with graph encodings, and a brand-new paper on grounding AI agents with graphs. 

Fourteen years after launch, the Knowledge Graph is more central to Google’s stack than ever, with new applications in AI Overviews, AI Mode, and agentic search.

For years, the SEO industry has responded to traffic volatility by treating content like an arms race. What Lily Ray calls the Vicious Cycle of SEO: spammers generate spam, Google releases spam updates, producing the “Mount AI” pattern we’ve all become familiar with. Search Engine Land’s trajectory points to a more sustainable alternative. Instead of writing more, declare more.

For smaller publishers, content clusters, reputation management, and page-by-page semantic SEO may be enough. But for publishers buying their digital ink by the barrel, taking control of your context infrastructure is probably the only viable play. That means connecting your content to the Knowledge Graph and routing your internal links through the topics that matter most. 

The publishers who declare their authority will keep being cited. The publishers hoping to be inferred will be left in the noise. The choice is yours.