Okay, I need to talk about something that's going to make me deeply unpopular with content managers everywhere. And honestly? I'm kind of looking forward to it.

The entire industry's response to AI search has been, essentially, "publish more stuff." More topic clusters. More keyword gaps filled. More pages indexed. It's the content equivalent of speaking louder when someone doesn't understand you — technically a strategy, but not the one you think it is.

A leaked set of Google ranking signals suggests the opposite might be true. That the most valuable discipline in AI search isn't knowing what to publish. It's knowing what to leave out.

I realise I'm writing a newsletter telling you to produce less content. The irony is not lost on me. (My therapist would call this "self-sabotaging behaviour." I call it "journalism.")

Here's what happened. Someone — probably someone very smart, probably at a conference with good coffee — said "cover your subject comprehensively." Solid advice. Hard to argue with. This newsletter has made that exact case more than once.

But somewhere between the conference stage and the content calendar, "cover your subject comprehensively" mutated into "publish more." Those are not the same instruction. The difference is roughly the difference between being well-read and just owning a lot of books. (We all know someone with an impressive bookshelf and suspiciously uncreased spines.)

The emerging evidence from how search engines actually model websites suggests that volume is close to irrelevant — and that content drifting away from what your site is genuinely about may be quietly working against you. Not just failing to help. Actively hurting.

The lever everyone's reaching for is the wrong one. The more consequential lever is restraint. Which is, admittedly, the least sexy word in marketing.

Let's get specific about what actually matters here, because "topical authority" gets thrown around a lot without much precision.

Topical authority is a search engine or AI system recognising that your site is the most trustworthy source on a particular subject. Not for one keyword. For the entire neighbourhood of related questions surrounding it. A law firm that's built genuine depth on employment law. A SaaS company that owns the conversation around workflow automation. A local physio practice that's become the definitive resource on sports injury recovery. That kind of ownership.

This is fundamentally different from what search marketers spent two decades building. Traditional domain authority — the old link-counting model — tells a search engine that lots of other sites vouch for you. Topical authority tells it something narrower and increasingly more useful: on this specific subject, you demonstrably know what you're talking about. You can have plenty of one and essentially none of the other.

The crucial point: these trust signals aren't scored for your site as a whole. They're evaluated relative to a subject. A B2B publisher can be a credible authority on supply chain logistics and carry zero weight on HR software, and the system is designed to tell the difference. You don't earn topical authority by being broadly reputable. You earn it by being consistently, verifiably expert in one place.

Here's where this stops being a nice-to-have and starts being the whole ballgame.

AI systems don't answer a question by running a single search. They decompose it — a mechanism known as query fan-out — into a spread of related sub-queries, run those in parallel, and assemble an answer from whichever sources surface most consistently across the whole set.

Think about what that means. A page that ranks for one narrow term appears in one slice of that spread and nowhere else. A site with genuine depth across the subject appears again and again, accumulating citation weight through sheer repetition. The unit that gets rewarded isn't the page. It's the topic.

This is true whether you're a professional services firm trying to get cited in ChatGPT, a SaaS company appearing in Perplexity, or a publisher showing up in Google's AI Overviews. The mechanism doesn't care about your industry. It cares about your depth.

For years, topical authority was something everyone agreed mattered but nobody could measure. Then, in 2024, a large set of internal Google ranking documentation became public, and two signals made it concrete.

The first: a site focus score — how concentrated your content is around a single core subject. The second: a site radius — how far any given piece of content sits from that centre.

In plain English: Google builds a model of what your entire site is about, and then measures how far each page wanders from that core. Off-topic content isn't neutral. It moves your centre of gravity, widens your radius, and blurs the picture the system is trying to build of what you are for.

Which is inconvenient news for anyone who's spent the last three years filling keyword gaps like their life depended on it. (So... basically all of us.)

Almost nobody set out to dilute their own authority. The trap is a chain of individually logical decisions.

A keyword tool surfaces a term with healthy search volume and low competition. It's adjacent to your core subject — not central, but not absurd. Someone asks: why not cover it? There's demand, the competition is soft, and more indexed pages has always felt like more surface area. Repeat that a few hundred times and you've built a site a search engine can no longer describe in a single sentence. It's like adding rooms to a house one at a time until you've accidentally built a hotel nobody asked to stay in.

What makes this worse: quality signals are now increasingly evaluated at the site level, not per page. Off-topic or thin content can drag down pages that would otherwise have performed well. Your best content is getting punished for keeping bad company.

The clearest evidence that concentration beats scale is what happens when the two collide directly.

Specialist sites with negligible domain strength routinely outrank enormous generalists within a narrow subject. A niche accounting blog outranking a Big Four firm's content hub for specific tax compliance queries. A solo consultant's site on data privacy outperforming a massive legal directory. A boutique agency's thought leadership getting cited by AI systems while the industry publication with ten times the traffic gets ignored.

The generalists have vastly more links, more traffic, more of every conventional strength. What they don't have is focus. On that one subject, the specialist owns a tighter, clearer topical identity, and in AI search, that's the variable that decides the outcome.

Size is not the moat it used to be. Concentration is.

The good news is you don't need to take this on faith. You can check whether your site has a focus problem in an afternoon.

The simplest version: list every page on your site and ask, for each one, could someone landing here for the first time tell what this site is about? Be honest. If a page about "top team-building activities" lives on a site about accounting software, you've found your drift.

For a more rigorous approach, AI tools can now analyse the semantic similarity of your content — essentially mapping every page and showing you which ones cluster tightly around your core subject and which ones have wandered off into the wilderness. The output is a hit list: pages reinforcing your authority on one side, pages diluting it on the other.

For most sites that have been publishing consistently for a few years, this exercise is uncomfortable. The drift is rarely dramatic — it's the accumulated residue of hundreds of reasonable decisions — and it usually turns out to be a larger share of the library than anyone expected.

Okay, so your site has probably drifted. Most have. Here's the practical playbook.

Before you publish anything new, run it through one question: would a first-time visitor landing on this page understand what this site is fundamentally about? If a piece of content would leave that visitor confused about your subject, it will do the same to a search engine, and it doesn't belong — however much traffic it might attract.

Audit your existing content. Go through your blog, your resource centre, your knowledge base. For each page, ask two things: is this topic genuinely within our core subject, or just loosely related? And do we have a real claim to expertise here, or did we just spot a keyword opportunity? A SaaS platform for project management writing about office interior design because "it's adjacent to productivity" is the textbook case of a decision that sounds reasonable in a planning meeting and reads as dilution to an algorithm.

Deal with the off-topic pages you find. You've got three options: consolidate them into something that is on-subject, redirect them to a relevant page, or remove them entirely. The instinct to leave them alone — they're not hurting anyone, they get a trickle of visits — is exactly the instinct the leaked signals argue against.

Reframe how you evaluate content ideas. Traffic potential, on its own, is no longer a sufficient reason to publish. Every piece needs to pass a second test: does this reinforce what we're known for? If it doesn't, it's not a free option. It has a cost, and the cost is paid by everything else on your site.

Define your one sentence. If a search engine had to describe your site in a single sentence, what would it say? If you can't answer that clearly, neither can the algorithm. That sentence becomes your editorial north star — everything you publish should make it more true, not less.

There's a genuine tension here, and it's worth naming rather than papering over.

The pressure to publish more is intensifying. AI makes content cheaper to produce, and the temptation to flood every conceivable sub-query grows by the week. The signals point the other way entirely. The durable advantage in AI search belongs not to whoever produces the most, but to whoever stays the most disciplined about what they refuse to produce — the sites narrow enough that a machine can describe them in a sentence, and confident enough to leave the adjacent, tempting, off-topic traffic to someone else.

That is a genuinely awkward thing to tell a content team measured on output. It is also, I suspect, the most valuable strategic insight a lot of organisations could absorb right now.

The question worth carrying into your next planning meeting isn't what else could we cover. It's what would we be better off never publishing at all.

And yes, I'm aware that's a strange position for someone who publishes a newsletter three times a week. Self-sabotage. My therapist is going to have a field day.

Jo Lambadjieva is an entrepreneur and AI expert in the e-commerce industry. She is the founder and CEO of Amazing Wave, an agency specializing in AI-driven solutions for e-commerce businesses. With over 13 years of experience in digital marketing, agency work, and e-commerce, Joanna has established herself as a thought leader in integrating AI technologies for business growth.

😍It was great, all I needed

🙂Good, not great

🙄It sucked