Cover everything, own nothing

At its Search On event this month, Google walked through a set of advances in how it understands language and content, and one of them quietly changes how you should think about broad topics. Google said it has applied neural networks to understand the subtopics within a broad interest, so that a broad search can surface a wider variety of content. The example it offered makes the idea concrete: a search for home exercise equipment can now be read as containing subtopics such as budget equipment, premium picks, or small-space ideas, and the results page can show a spread across those facets rather than ten pages all taking the same general angle. Google said it would begin rolling this out by the end of the year. It sat among other announcements at the same event — continued improvements in reading the content of pages, and the striking note that BERT, the language-understanding system used on a small slice of English queries when it launched a year earlier, is now used on almost all of them. The through-line across all of it is Google getting better at understanding meaning rather than matching words. But subtopics deserves singling out, because it does not just improve results quality in the abstract; it changes the shape of a broad results page, and with it the kind of page that can win a place there. The rest of this note is about that change and what it asks of you.

To see why this matters, look at what diversification does to a broad query. The old model was a single ranking: for a broad search, Google returned a list of pages competing to be the best overall page about the topic, and a comprehensive page that covered the whole subject solidly could earn a strong position by being thorough. Subtopics replaces that with something closer to a curated set. When Google understands that a broad interest contains distinct facets, it can assemble a results page that deliberately covers the range — a strong page for the budget angle, another for premium, another for small spaces — rather than stacking ten pages that all say roughly the same general thing. The page stops being one competition and becomes several, one per facet, with Google actively seeking variety across them rather than redundancy. That is a meaningful structural change. It means the question a results page is asking has shifted. It used to ask, in effect, which pages are most about this topic. It increasingly asks which page is the best on each distinct part of this topic, and how do I show a range. Those are different questions, and they reward different pages. The first rewards comprehensiveness; the second rewards distinctiveness on a facet. A strategy tuned for the first — cover the whole topic, thoroughly, in one place — is tuned for a competition that is quietly being replaced.

Here is the uncomfortable consequence for a lot of broad content, and it is worth stating plainly. When a results page is assembled to show variety across subtopics, it has no reason to include a page that is merely adequate across all of them. Run the logic. Each diversified slot represents a facet, and each slot goes to the page that is most clearly the best on that facet. A page that covers every subtopic at moderate depth is, by its own design, not the best on any single facet — it cannot be, because its attention is spread across all of them while its competitors poured everything into one. So it loses the budget slot to the page that is genuinely the best on budget, the premium slot to the premium specialist, and so on down the line, winning none of them. The breadth that was an asset under the old single-ranking model becomes the precise reason the page is now invisible: it is structurally redundant on a page that is curated to avoid redundancy. This is not Google penalising broad content; there is no penalty here at all. It is just what happens to undifferentiated work when the surrounding results are chosen for range. The generic page is exactly the sameness the diversified page is built to exclude. And the lesson cuts to something the AC Group has argued for years: authority is earned on something specific. Being a passable voice on everything was always weaker than being the recognised best on something; subtopics simply makes the SERP enforce that truth. The distinct page earns its place; the general one is the duplication the page is designed to leave out.

The response is not to abandon broad topics but to change how you compete for them: stop trying to win a broad topic with one broad page, and start winning its subtopics with pages that each go genuinely deep. Treat the broad topic as a map of facets rather than a destination you rank for by being vaguely about it. Concretely, that means finding the real subtopics — and they are not hard to find, because Google and your audience already articulate them. Look at what the broad query returns, at the related searches and the questions people also ask, at the way results already cluster into angles like price tiers, use cases, skill levels, or constraints; those clusters are the subtopics, handed to you. Then cross them against where your genuine depth lies — the facets you can write about with specifics a generalist could not fake, where you have real experience or a defensible point of view — and choose those. Build a focused page for each chosen facet that is unmistakably the best answer to it, and let the broad topic become the structure linking them rather than a page competing on its own. The trap to avoid is reading subtopics as licence to mass-produce thin pages on every conceivable facet; that just spreads the same shallowness across more URLs and earns nothing, because the bar is depth and distinctiveness, not mechanical coverage. A few facets owned with real authority will beat a dozen touched lightly, every time the results page is built for variety. Mapping a broad topic into the subtopics you can authentically be the best at, and earning a place on each, is the discipline the AC Group has worked by for 27 years.