Long-tail keywords: how to find them

Long tail keywords are specific searches that each get few searches a month, like "waterproof trail running shoes for wide feet" instead of "running shoes". You find them in Google autocomplete, People Also Ask, your own Search Console queries, forums, support tickets and the questions customers type into AI assistants. Most of them don't need their own page. They belong as sections, headings and sentences on a page that covers the broader topic.

What are long tail keywords?

A long tail keyword is a query with a narrow meaning and low volume per term. The name comes from the demand curve. A few head terms get huge volume, and behind them sits a long tail of thousands of rarer queries that together add up to more traffic than the head.

Word count alone doesn't decide it. A two-word query like "invisalign retainer" can be long tail too, if few people search it and it means one thing. And a five-word query like "how to lose weight fast" is a head term in disguise, with heavy competition and vague intent.

The test that works is simple. If you can tell exactly what the searcher wants and one paragraph could answer it, it's long tail.

The tail is deeper than any tool shows. In its 2019 BERT announcement, Google said 15 percent of the queries it sees every day are new to it. No keyword list contains those, so write pages that answer real questions instead of chasing exact strings.

Why long tail keywords convert and are easier to rank for

Specific queries come from people further along in a decision. Someone searching "crm" might be a student writing an essay. Someone searching "crm for a 3 person roofing company with quickbooks sync" is shopping, and the query tells you what to put on the page.

They're easier because fewer sites answer the exact question, and big sites often answer it with a generic page that a precise one can beat. How much easier depends on who already ranks, which our post on low-competition keywords covers.

The cost is volume. Each term brings a handful of visits, so long tail work pays off in aggregate. One page that answers twenty related specific questions well can outdraw a page aimed at one head term it never ranks for.

How to find long tail keywords

Start from a seed term and collect the specific versions people use.

Google autocomplete

Type your seed into Google and read the suggestions. Then add a letter, or "for", "vs", "without", "with" after the seed and read again. Autocomplete reflects what people type, so it's the fastest raw source. It shows no volume. Check that separately with a keyword search volume source before you rank priorities.

People Also Ask

The question boxes in Google results are long tail queries in question form, and each one you expand loads more. Our guide to People Also Ask covers how to mine them and what to do with the answers.

Search Console queries

This is the source most people skip, and it's the best one, because it shows long tail queries your site already gets impressions for. Open Performance, add a query filter, choose "Custom (regex)" and paste a pattern. This one keeps queries of five words or more:

^\S+(\s+\S+){4,}$

Sort by impressions. Queries with impressions but almost no clicks, where you rank on page two, are your easiest wins.

Forums and Reddit

Read the thread titles in the communities your buyers use. People describe problems with the brand names and constraints that keyword tools flatten.

Support tickets and sales calls

Your inbox holds questions nobody else has data on. Export a month of tickets, pull the first sentence of each and group the repeats. Those are long tail keywords with proven demand from your own customers.

AI prompts

A ChatGPT prompt carries context a search box never gets, so every prompt is long tail. You can't see other people's prompts, but you can approximate them, as our post on questions people ask ChatGPT explains. AI search also breaks one prompt into several narrower searches, described in query fan-out, so pages that answer specific sub-questions get more chances to be cited.

Does each long tail keyword need its own page?

Usually no. If two queries would be answered by the same page, write one page. "Best crm for roofers", "roofing crm with quickbooks" and "crm for small roofing company" are one page with sections, not three thin pages competing with each other.

Give a long tail keyword its own page only when the answer is different enough that it wouldn't fit as a section. A comparison, a separate product, a location or a how-to with its own steps usually qualifies. A rephrasing does not.

Splitting too far causes keyword cannibalization, where your own pages compete for the same query and none ranks well. Group your list by meaning first, with a keyword clustering pass, then decide pages from the groups.

Find long tail keyword ideas from autocomplete

Our Keyword Ideas tool runs your seed through Google autocomplete for the market you pick and returns your seed plus up to 10 suggestions. For each one it shows the word count, the words it shares with your seed, the new modifiers it adds and whether it's shaped like a question. That last part flags the ones to answer as a heading.

It costs 15 credits per run. It doesn't show search volume, difficulty or cost per click, and it doesn't pull People Also Ask or related searches. It also can't see your Search Console data. Use it to fill the top of the list quickly, then check volume and group the results before you write.