Topical authority is not domain authority
These get used interchangeably in pitches and they describe opposite things. Domain authority is a third-party estimate of your link strength across the whole site. Topical authority is a description of how thoroughly you cover one subject — and it's the reason a 40-page site about one narrow thing routinely outranks a 4,000-page brand that mentioned it twice.
The mechanism isn't mystical. A site that answers forty questions about dental implants has more chances to match a query, more relevant internal links pointing at each page, more repeat visits from people researching the same decision, and a clearer subject signal than a general health portal with one implants article. None of that requires a hidden authority score to explain.
| Topical authority | Domain authority | |
|---|---|---|
| What it describes | How completely you cover one subject | A third-party guess at overall link strength |
| Who calculates it | Nobody — it's a description, not a score | Moz, Ahrefs and others, from their own link indexes |
| How you move it | Publish and interlink across the whole topic | Earn links from more unique domains |
| What it explains | Why small niche sites beat big general ones | Roughly how hard a site is to outrank in general |
| How it fails you | Confused with page count | Treated as a target instead of a filter |
Turning it into a countable plan
"Build topical authority" is not a brief. "We cover 34 of the 71 questions a buyer asks about this, here are the 37 we don't" is. The conversion takes a day.
Build the question map
Collect every real question in the topic from sources that contain actual humans, not from a keyword tool alone.
- Search Console queries you already get impressions for — including the embarrassing ones you don't have a page for.
- People Also Ask boxes, expanded two levels deep, for your five head terms.
- Your sales calls. The objection someone raised on Tuesday is a page.
- Reddit, Quora and the WhatsApp groups your buyers are in.
- Your competitors' navigation — not their blog list, their menu. That's their considered view of the topic.
Score your coverage
Dedupe the list into distinct questions — sixty to eighty is typical for a serious commercial topic. Assign each one to a page, allowing that a good page can genuinely answer three or four related questions. Then divide: questions with a page that actually answers them, over total questions.
That percentage is your coverage. Track it monthly. It's the only honest way to answer "how much content do we still need", and it stops the argument about whether to publish four posts or twelve — the map tells you when you're done.
Wire it together
A map with no links is a folder, not a body of work. Every page links up to its hub, the hub links down to every page, and pages link sideways only where a reader would genuinely want the other one. See topic cluster for the pattern in detail.
Use the destination page's own subject as the anchor text, not the head term. Forty internal links all reading "dental implants" tell Google nothing about which page owns which question.
How long it takes, and the honest caveat
Individual long-tail pages can pick up impressions within weeks — they're competing against thin, ageing content in most Indian categories. The compounding effect, where new pages in the same topic start ranking faster than your first ones did, shows up across quarters rather than months. On a competitive subject, expect six to twelve months before a topic feels like yours, and longer if your site has no link profile to speak of.
The caveat worth stating plainly: Google has never confirmed a topical authority score, and some experienced SEOs argue the whole concept is a folk explanation for ordinary relevance and internal linking effects. That argument is worth reading — we've written our own view in does topical authority actually exist.
It doesn't much matter who's right. The plan the theory produces — map the real questions, answer them properly, link them together, measure coverage — is the correct plan under either explanation. That's the test of a useful model: it survives being wrong about the mechanism.