What a long tail keyword is, and how to tell one worth a page

A long tail keyword is a search phrase that few people type but that describes exactly what the searcher wants. It sits at the thin end of search demand: a handful of searches a month each, intent that is already obvious, and far fewer pages written to answer it. The test for whether a phrase is long tail is not how many words it has. The test is how much of the searcher's situation it pins down. "Crm" tells you nothing about the person. "Crm for a two person estate agency" tells you the industry, the size, the budget range and the objection you need to answer, and it does it in six words.

That distinction matters because most keyword lists get sorted by length and worked top to bottom, which puts the effort in the wrong place. This page gives you the specificity test, the four shapes a tail phrase takes, what the volume figure in your tool is really saying, and how to decide whether a phrase deserves a page.

The test: how much does the phrase decide for you?

Read a phrase and ask what you would still have to guess if you had to write the answer. The fewer the guesses, the further down the tail it sits.

Phrase	Words	What you still have to guess	Where it sits
insurance	1	Everything: the product, the buyer, the country, the intent	Head
travel insurance	2	The traveller, the trip, the cover, whether they are comparing or buying	Head
travel insurance for pre existing conditions	6	Which condition, and the destination	Middle
travel insurance for a diabetic over 70 going to spain	10	Almost nothing	Tail
dpdp consent manager	3	Almost nothing, because the term itself is precise	Tail

Read the last two rows together. One is ten words and one is three, and both are long tail, because the three word phrase is a precise term used by people who already know what they need. In ordinary language, length is a side effect of specificity. It is not the thing itself.

Why the word count rule gets taught anyway

The rule survives because it is usually right by accident: vague people type two words and specific people type a sentence. It fails in two places that cost real money. Short technical terms get dismissed as head terms and left unwritten, and that is where a specialist business wins quietly. Long vague phrases get treated as easy wins, and "what is the best marketing strategy for a small business" is ten words of nothing: every page in the category is an answer to it.

The four shapes a tail phrase takes

• The question. "Can i claim input tax on a laptop bought for my business". Typed or spoken as a full sentence, and answerable in one paragraph.
• The qualified category. "Noise cancelling headphones for a small head". A head term plus the constraint that makes this buyer's choice hard.
• The named comparison. "Zoho books vs tally for a chartered accountant". Two options and the context that decides between them.
• The exact object. "Chuck size for a bosch gsb 550", "wiper linkage part number 2016 swift". Nobody types this unless the thing is in front of them.

Each shape wants a different page. A question wants the direct answer in the first two lines. A qualified category wants a short list with the constraint actually applied. A comparison wants a table and a recommendation. An exact object wants the specification and the next step to buy it or fix it.

The volume figure, and what it is not telling you

Every keyword tool shows a low number or a zero against a genuine tail phrase, and both readings are easy to misread. Google's own Keyword Planner reports volumes in rounded bands and groups close variants together, so the figure beside one phrasing is often the figure for a cluster of them. A zero means the phrase sat below the reporting threshold, which for a tail phrase is the normal condition rather than a verdict on it.

Your own data settles it. The queries table in the Search Console performance report lists the phrases people actually used to reach you, including the ones no tool will quote a number for. Sort by impressions and read the rows as sentences rather than as keywords. A phrase with eleven impressions and one click is a real phrase with a weak answer behind it, which makes it the most useful row on the report.

Three checks before a phrase earns a page

Specificity makes a phrase long tail. It does not make it worth your afternoon. Run three checks.

• Does it need a different answer? If the honest answer matches one you have already published, it is the same question in different clothes and it belongs in the page you already have.
• Does the searcher buy what you sell? "How to fix a dripping tap yourself" and "emergency plumber open sunday" are both tail phrases. One brings a reader and one brings a job.
• Can you be the best answer on the page? Tail questions are frequently answered badly by everybody, and that is the opening. Where a manufacturer, a regulator or a marketplace already answers it definitively, your effort earns more somewhere else.

How to target one properly

One page, one question, answered at the top. Put the direct answer in the first two sentences, in the words the searcher used, then earn the rest of the page: the reason behind the answer, the cases where it changes, and what to do next. Use the natural variants inside the writing rather than repeating the phrase. Title it with the question or the thing, link to it from the broader page on the subject, and link out to the deeper pages a satisfied reader wants next.

One note on scope, because this is where sites go wrong at exactly this point: a page for every phrasing of one question leaves three pages competing for the same answer and none of them winning it. Group the phrasings and write once. The fuller treatment of the distribution, and how to build a set of pages across it, is in our guide to the long tail of search.

Measuring a tail page without a rank tracker

A position for a phrase with eleven searches a month moves on noise, so measure the page rather than the phrase. Filter the performance report to that page's URL and watch the number of distinct queries it collects, alongside impressions and clicks. A healthy tail page gains queries you never wrote down, and that list is your next brief.

Word counts, zero volume figures and one page per phrase

How many words does a keyword need before it counts as long tail?

There is no threshold, and a word count is the wrong instrument. Three precise words are long tail and ten vague ones are not. Ask what the phrase decides for the person writing the answer, and when little is left to guess, it is a tail phrase.

Are zero volume keywords worth writing for?

Yes, where the zero is a reporting threshold rather than an absence of demand, and that is usually what it is. A phrase that turns up in your sales calls, your site search or your Search Console queries has demand whatever an estimate says. The ones to leave alone are the phrases nobody has typed anywhere you can observe.

Do long tail keywords convert better than head terms?

Usually, and the reason is in the phrase rather than in the tactic. Somebody who names the constraint, the size or the part number has already made most of the decision. A head term collects people at every stage at once.

Should each long tail phrase get its own page?

Only where it needs its own answer. Phrasings that share an answer share a page, and the variants find their place in the headings, the examples and the closing questions. Two phrases that need different answers were two questions, and they deserve two pages.

Where do tail phrases come from if the tools will not show them?

From the places where people write in their own words: the Search Console queries report, your site search log, autocomplete, support tickets, and the first sentence of a sales call before anybody tidies it up. Our pages on finding keywords from the sources you already own and what a keyword is in SEO cover collection.

The tail is also where people now ask an assistant

Look again at what a tail phrase is. "Travel insurance for a diabetic over 70 going to spain" is already a sentence, and a sentence is what people type into ChatGPT, Gemini or Perplexity. The specific end of the curve moved first, because a full question is exactly what those tools were built to take. What comes back is one paragraph and two or three named sources, and the page that would have ranked third for that phrase is not in the room.

So the tail questions you are writing for are worth asking twice, once of Google and once of an assistant, in the same words. The second answer shows you which sources are being named on your most specific questions, and whether the page you just published is among them. Do it by hand on your five best tail questions this week. For keeping the record over time, and for seeing it move after you publish, there is a tool for it at AI Knows Us.