The Long Tail Is Where the Answers Are

Every keyword tool sorts by volume, so every content plan starts at the top: the few hundred terms with big numbers next to them. Those terms are real, and they are the hardest thing on the internet to rank for. What the sorted list hides is the shape of the rest — and the rest is where most demand, most AI answers, and most of the citations now sit.

This report puts the shape of demand next to the newest data on where AI answers appear and what they cite. The conclusion is not new to us; it is the strategy we have run for more than a decade. What is new is that answer engines have made it the only strategy that scales.

1. The shape of demand

Ahrefs’ U.S. keyword database holds 2.3 billion keywords with fewer than ten searches a month — almost 93% of the database, per its May 2026 update (the 2021 edition of the same analysis put it at 94.74%). At the other end, roughly 18,000 keywords see more than 100,000 monthly searches. Fifteen percent of the queries Google sees on a given day have never been searched before, a figure Google has restated repeatedly since 2013.

Source: Ahrefs, Long-tail Keywords (Soulo, updated May 27, 2026).

Two consequences follow. First, there is no way to plan for the long tail one keyword at a time; the individual queries are too small and too numerous, and a seventh of them did not exist yesterday. Second, the pages that serve it are worth having anyway, because in aggregate they are most of search. Ahrefs’ separate finding that 96.55% of web pages get no Google traffic is the mirror image: most pages are not specific enough to match any of those two billion small questions.

2. AI answers live at the long end

The newest evidence changes the stakes. Ahrefs’ November 2025 analysis of 146 million results pages found an AI Overview on 20.5% of them — and the rate climbs steeply with the length of the query.

Source: Ahrefs, AI Overview triggers (Law, November 10, 2025); 146 million SERPs.

Query type matters as much as length: 57.9% of question queries and 59.8% of “why” queries produced an Overview, against 6.3% of news queries; definitions 47.3%. By intent, 99.9% of Overview-triggering keywords were informational. By topic, science (43.6%) and health (43.0%) led; shopping (3.2%) and real estate (5.8%) trailed. Pew’s behavioral panel saw the same shape from the user’s side in March 2025: 8% of one- or two-word searches produced an AI summary, 53% of ten-plus-word searches, and 60% of question-format searches.

The long tail was always where people asked specific questions. It is now where they get answered without a click — unless the answer needs a source.

3. Where the citations come from

That “unless” is the whole opportunity. In our report on citations, we covered Ahrefs’ March 2026 finding across 863,000 keywords: only 37.9% of pages cited in AI Overviews rank in the top ten for the query, 31.2% rank between 11 and 100, and 31.0% do not rank in the top hundred at all. Ahrefs attributes part of the shift to Google’s January 2026 move of AI Overviews to Gemini 3, with better handling of long-tail questions.

Read that against the demand curve. Nearly a third of the pages answer engines draw on are pages a rankings report would never show — pages that exist for a specific question and are retrieved because they answer it, not because they rank. A resource that already holds a page for every entity, cohort, place, and comparison in a topic has a candidate for every phrasing of every question. A resource with one strong page per head term has a candidate for the head term.

4. Winning the head from below

None of this means abandoning the head terms. It means the head terms are won from the base, and the data on how pages come to rank says why.

Source: Ahrefs, How long does it take to rank in Google? (Stox, May 2025); 1.3 million keywords, 1 million new URLs.

The average page in position one is five years old; it was two in 2017. Rankings are an accumulation — of links, of engagement, of the trust that a stable address earns over time — and accumulation happens fastest where competition is thinnest. A long-tail page about one program at one school faces little competition, wins quickly, earns a link from the department it describes, and passes that authority up through the cluster: to the school’s page, to the “best programs in the state” ranking, to the head term. Ahrefs’ own correlation work finds referring domains the strongest predictor of traffic, and most of those domains link to something specific.

The pattern in one vertical, from our own keyword work: at the head, “nursing school”; in the mid-tail, “online nursing degree” and “nursing school rankings”; in the long tail, “best nursing schools in California for undergraduates,” “nursing informatics degree online,” “[college] school of nursing graduation rate.” The site that covers the third tier completely — every school, every state, every specialization — is the site that eventually holds the first, and it gets there by being the source for ten thousand specific questions on the way.

5. The economics: why it has to be templates

The reason most publishers do not do this is not that they disagree. It is that the long tail is unaffordable one article at a time. Two billion small questions do not have a budget line, and a writer producing one page per school per program would need centuries.

The only way to cover the long tail is to cover it structurally: one well-designed template, rendered against a knowledge graph, for every entity in the category. That is how one flagship site came to publish thousands of distinct rankings and a partner network of ten sites came to hold 1.8 million distinct ranking lists across more than a hundred templates — and how the team behind them did roughly ten times the work with half the people between 2019 and 2021. It is also, not incidentally, how the pages stay current: when the data refreshes, every page in the tail refreshes with it, which is the freshness that answer engines have been measured to prefer.

Templates do not make the writing worse. They change what the writer writes: the pattern, the comparison logic, the insight rules, once — reviewed once — instead of the ten-thousandth variation of a sentence whose only difference is the data.

Methodology and caveats

Demand figures are from Ahrefs’ U.S. keyword database as described in its long-tail keywords guide (updated May 2026); Ahrefs’ database is large but not a census of queries, and volumes are estimates. The 15% new-query figure is Google’s own, restated most recently at Search Central Live in 2025. AI Overview trigger rates are from Ahrefs’ analysis of 146 million SERPs (November 2025); prevalence has changed since and varies by market. Citation-rank overlap is from Ahrefs’ March 2026 study (863,000 keywords). Page-age and time-to-rank figures are from Ahrefs’ May 2025 study of 1.3 million keywords and 1 million new URLs; the 2017 comparison figures are Ahrefs’ own earlier edition. Our figures — ranking-list counts, templates, and team productivity — are from internal reporting compiled in 2022; the keyword tiers in section 4 are illustrative of real clusters we operate, not a ranked list.

What to do about it

1. Plan by entity, not by keyword. List the things in your vertical — the schools, providers, products, places, employers — and the attributes people ask about. That grid is the long tail, and it is finite even when the queries are not.
2. Build one excellent template per page type and render it for everything. Review the pattern once; let the data do the variation.
3. Give every long-tail page a real answer — the number, the comparison, the source, the date — because those are the pages answer engines retrieve when nothing in the top ten fits the question.
4. Link upward deliberately. Every specific page should point to the broader page it supports. Authority moves along links; design the paths.
5. Measure the tail in aggregate. Individually, long-tail pages look like nothing. Judge them by template: clicks and citations across all pages of a type, not per page.
6. Be patient about the head, impatient about the tail. The long-tail pages win in weeks; the head term takes the years that Ahrefs’ age data describes. Both are on schedule.

Sources

1. Ahrefs — Long-tail Keywords: What They Are and How to Get Search Traffic From Them (Soulo, updated May 27, 2026) — Almost 93% of U.S. keywords under 10 searches/month; ≈2.3 billion keywords
2. Search Engine Land — Google reaffirms 15% of searches are new, never been searched before — Google's own figure, restated since 2013
3. Ahrefs — 96.55% of Content Gets No Traffic From Google (Soulo, December 2023)
4. Ahrefs — What Triggers AI Overviews? (Law, November 10, 2025) — 146 million SERPs; rates by query length, type, intent, and industry
5. Pew Research Center — Google users are less likely to click on links when an AI summary appears in the results (July 22, 2025)
6. Ahrefs — Google AI Overview citations and top-10 rankings (Linehan, March 2, 2026) — 863,000 keywords, 4 million cited URLs
7. Ahrefs — How long does it take to rank in Google? (Stox, May 15, 2025)
8. Ahrefs — Do AI assistants prefer to cite fresh content? (Law, July 28, 2025)
9. Media Factual — internal reporting on ranking lists, templates, and team productivity (compiled 2022)