The demand curve, and where the money actually sits
Plot every query against how often it's typed and you get a very short, very tall head and a tail that goes on almost forever. A handful of queries get enormous volume; millions get a handful each. In aggregate the tail is the majority of all search — a large share of daily Google queries have never been searched before, which is why it never runs out.
The point isn't that the tail is big. It's that the tail is where people describe their actual situation. "Seo agency" tells you nothing about who's typing. Add four words and you know their business model, market and stage.
One worked example: the same business, two queries
Take a Bangalore D2C skincare brand looking for an agency — the two ends of the same demand curve.
| "seo agency" | "seo agency for d2c brand in bangalore" | |
|---|---|---|
| Monthly volume | Thousands across India | Often reported as 0–10 by tools. Real demand exists; the tool just can't measure it. |
| Who's typing | Students, job-seekers, agency staff, curious founders | A founder or marketing lead with a budget and a shortlist |
| What's ranking | National directories, listicles, agencies with years of link equity | A handful of agency pages, some of them not even D2C-focused |
| Realistic time to rank | 12+ months, with sustained link work, if at all | Weeks to a few months on a decent domain |
| What a click is worth | Very little. Most of it never becomes an enquiry. | High. The query is a qualification form the searcher filled in voluntarily. |
Specificity is the point, not length
"How to do search engine optimisation for my business website" is eleven words and still head intent — vague, broad, competing with every beginner's guide on the internet. "Ecommerce seo cost india" is four words and squarely long-tail: rare, specific, typed by someone with a decision to make.
So drop the word-count heuristic. The test is whether the query narrows the field: does it name an industry, city, business model, price bracket, constraint or comparison? Each cuts both competition and ambiguity.
Specificity also buys a matchable format. Broad queries have mixed search intent and Google hedges. Specific queries have one obvious answer shape, so you know what to build.
- Industry: "seo for dental clinics" instead of "local seo"
- Geography: "seo agency in kochi" instead of "seo agency"
- Business model: "b2b saas seo" instead of "saas marketing"
- Constraint: "seo for a site with no blog", "cheapest seo package india"
- Comparison: "agency vs freelancer for seo" — see agency vs freelancer
Where to find yours, without paying for a tool
The best source is your own Search Console, because it lists real queries real people used to reach you — not modelled estimates. Everything else is a supplement.
- Search Console, Queries tab, six-month window. Sort by impressions and read past the first fifty rows. The long-tail lives at the bottom of that list, and about a third of it will be phrasing you'd never have written yourself.
- Google autocomplete. Type your head term plus each letter, a to z. Then type it as a question — "how much does", "which", "is". Free, live, and drawn from actual query logs.
- People Also Ask. Expand two levels deep. Each expansion pulls fresh questions clustered around the same topic.
- Your inbox and sales calls. The exact words a prospect uses in a WhatsApp enquiry are long-tail queries written by hand. This is the source everyone has and nobody mines.
- Reddit, Quora and industry forums. People ask questions there in full sentences, which is precisely the shape of a long-tail query.
Long-tail is how AI and voice search work
Nobody speaks in head terms. Ask an assistant out loud and you get a full sentence with context attached — "which seo agency in bangalore works with skincare brands". Typed into a chat assistant it's longer still.
Same shape as a long-tail query, which makes pages built for specific questions the easiest for these systems to use. An answer engine looking for a passage to cite wants a page that addresses the exact question, not a broad guide that covers it in paragraph nine.
So structure pages with each question in its own clearly headed, self-contained answer. Good for readers, good for featured snippets, and exactly what a citation-hungry model needs — more in answer engine optimization.