A team publishes forty posts a quarter. Traffic is flat, nothing gets cited in AI answers, and the content calendar has quietly become a treadmill nobody can step off. I see this pattern constantly, and the fix is almost never “publish more.” Much of the Cory Maki SEO approach starts from one observation: search engines and AI assistants both reward depth on a subject, not volume across subjects. A site that covers one topic exhaustively beats a site that covers thirty topics adequately — and in AI search, the gap is getting wider, not smaller.
What topical authority actually means
Topical authority isn’t a score you can look up. It’s the cumulative result of a site covering a defined subject area so completely that both crawlers and language models treat it as a reliable source on that subject. Practically, it shows up as three things: you rank for terms you never explicitly targeted, your pages get pulled into Google AI Overviews (the AI-generated summaries that sit above traditional results), and tools like ChatGPT and Perplexity cite you when someone asks a question inside your subject area.
The volume approach assumes each post is a lottery ticket — more tickets, more chances. That logic broke down years ago in classic SEO and it breaks down harder in AI search, because generative systems don’t retrieve sites, they retrieve passages. A thin post that mentions your topic once has nothing worth retrieving. A cluster of pages that answers every adjacent question about that topic has dozens of extractable passages, all reinforcing each other.
Why depth matters more in the AI era
Over a decade in reputation and search, I’ve watched the surface area of “winning” change. It used to be ten blue links. Now a meaningful share of questions get answered before anyone clicks, which means the goal shifts from ranking to being the source the answer is built from. Citations beat rankings in AI search — being quoted in an answer about your category is worth more than position four on a page nobody scrolls.
That shift punishes shallow breadth in a specific way. When a model assembles an answer, it draws from retrieved chunks of text and weighs how consistently a source handles the entity in question. If your domain has one post loosely touching a topic and twenty posts about unrelated things, there’s no consistent signal. If your domain has a hub page plus fifteen supporting pages that all handle the same entity with the same terminology, you become the obvious source to pull from. This is the same reasoning behind the ARC Method and AI-citation frameworks I’ve built out — earning a citation is a function of coverage, clarity and corroboration, not of publishing cadence.
How it works: the mechanism, with an example
Say a SaaS company sells security compliance software. The volume approach produces posts about productivity tips, remote work culture, a roundup of AI tools, and one post titled “What is SOC 2?” Nothing connects. The depth approach picks the boundary — SOC 2 for early-stage software companies — and covers the entire question set inside it:
- What SOC 2 is, and who actually needs it
- Type I versus Type II, and when each makes sense
- What a readiness assessment involves
- How evidence collection works day to day
- How to choose an auditor
- SOC 2 compared to ISO 27001
- What happens when you fail a control
- What buyers actually ask for in security reviews
Each page answers one real question completely, links to the others with descriptive anchor text, and uses consistent terminology throughout. Now when a founder asks an AI assistant how long SOC 2 readiness takes, the retrieval layer finds a passage that answers exactly that — on a domain that also covers the eight questions surrounding it. Corroboration inside your own site is a real signal. So is the absence of contradiction.
The same logic applies to service businesses and public figures. In my work with clients, the sites that get cited are the ones where a reader — or a model — can trace a complete argument across several pages without hitting a dead end. That’s also why the SEO fundamentals that still matter haven’t gone anywhere: crawlability, clean internal linking and clear on-page structure are what make depth legible to machines in the first place.
The Cory Maki SEO approach to building depth
The way I think about this: pick a subject you can plausibly own, then refuse to publish anything outside it until you’ve covered it. Here’s the working sequence.
1. Define the boundary narrowly
“Marketing” is not a topic. “Compliance content for pre-Series A SaaS” is. The narrower the boundary, the faster you reach the coverage threshold where results compound. You can widen later; nobody has ever succeeded by starting wide.
2. Map the real question set
Not just keyword tools. Pull questions from sales calls, support tickets, community forums and Reddit threads, where people ask things in their own words and get corrected by strangers. That’s where Reddit authority intersects with AI search — the phrasing people actually use is the phrasing models are trained and prompted with.
3. Build a hub and commit to the spokes
One comprehensive pillar page that defines the subject, plus dedicated pages for each sub-question. One page per intent — resist the urge to write three overlapping posts on the same question, which splits signals and confuses retrieval.
4. Structure every page for extraction
Clarity and structure make content citable. Answer the headline question in the first two sentences. Use headers that mirror how people ask. Define jargon on first use. Add a comparison table or a short list where the content warrants one. If a model has to infer your answer from three scattered paragraphs, it will pick a source that stated it plainly.
5. Link internally with descriptive anchors
Internal links are how you tell a crawler which pages belong to which subject, and how you pass context between them. Use anchor text that names the concept, not “read more.” This is quiet, unglamorous work and it does more for topical authority than most external link campaigns.
6. Refresh before you expand
Once coverage is complete, the highest-return move is updating what exists — new examples, corrected claims, sharper answers. One thing that consistently works: auditing the cluster quarterly and pruning or merging anything thin. Systems and automation scale quality here; a repeatable audit beats a burst of enthusiasm.
What depth does for reputation
There’s a second-order effect worth naming. When a model summarizes your company or your name, it’s synthesizing whatever it can find. A deep, coherent body of work on one subject gives it something specific and favorable to say. A scattered archive gives it nothing, and it fills the gap from elsewhere — reviews, forums, news, whatever ranks. That overlap between search strategy and AI reputation management is where most of my client work now lives, and it’s why I treat content depth as a reputation asset rather than a traffic tactic. Reputation is earned, not bought — and depth is how you earn it in a format machines can read.
If you’re new here, the introduction to what this site covers gives the broader map, and I go deeper on Generative Engine Optimization separately.
The durable principle
Search interfaces will keep changing. What doesn’t change is that both readers and retrieval systems are trying to answer the same question: who actually knows this subject? Volume can’t answer that. Complete, well-structured, internally connected coverage of a defined topic can — and it keeps paying off whether the answer is delivered as ten blue links, an AI Overview, or something that doesn’t exist yet. Show up where the answers are formed, cover your subject like you intend to own it, and let the breadth come later.