Long tail keywords are search queries that each get only a small number of searches, usually because they describe a very specific need, like “crate training a puppy that cries at night” instead of “dog training”. Here is how to find long tail keywords: start from a broad topic and dig into the places where real people phrase their exact problem, such as Google autocomplete, People Also Ask, your own Search Console queries, forums like Reddit, your site’s internal search, and the question and word-count filters in any keyword tool. Then group the phrases so each page answers one need. That last step matters as much as the finding.
Most guides list the sources and stop there. You end up with 400 phrases and no idea which ones deserve their own page, which is how sites end up with ten thin posts competing for the same search. I’ll cover both halves, with one worked example that runs from seed to finished page plan, as part of the expansion step in my keyword research process.
What Actually Makes a Keyword Long Tail?
Specificity, not length. Ahrefs defines long tail keywords as “search queries that get a small number of searches per month”, and its long tail keywords study makes the point directly: some one-word keywords get fewer than 100 searches a month, while some five-word phrases get hundreds of thousands. So word count is a clue. It isn’t the definition.
I think about it this way. A long tail query is narrow enough that only a small slice of people type it, and narrow enough that you can tell exactly what they want. “Running shoes” could mean anything. “Running shoes for flat feet on a treadmill” tells you the problem, the context and roughly where the person sits in their buying journey.
The tail is also enormous. In the same study, Ahrefs reports that keywords with fewer than 10 searches a month make up almost 93% of its U.S. keyword database. That’s the reason small sites win here first. Big brands don’t bother writing for queries this narrow, and the pages that do write for them often answer the question badly.
Three Things That Change Where You Look
Before picking a method, I check three things. Each one moves the best source up or down the list.
- Does your site already get impressions? If yes, your Search Console queries are the best long tail source you’ll ever have, because they’re real and they’re yours. If not, you rely on Google’s suggestions and community sites.
- Do you have a paid keyword tool? Paid tools find long tails faster and add volume estimates. They also miss a lot of the very rare phrases, so they never replace the free sources.
- Is your topic a product, a service or a subject? Product and service topics produce buying-intent long tails (“price”, “near me”, “for small kitchens”). Subjects produce question long tails (“why”, “how often”, “is it safe to”).
How to Find Long Tail Keywords: 7 Methods
These run roughly from fastest to slowest. You don’t need all seven on every project. For a small site, the first four usually give me more than enough.
1. Google Autocomplete
Type your seed into Google and read the predictions. Google’s help page says autocomplete predictions “reflect real searches that have been done on Google”, which makes them one of the most honest free sources there is.
The trick is to keep typing. Don’t stop at the seed. Add a letter after the seed (“crate training a”, “crate training b”), then a question word in front (“how crate training”, “why crate training”). Each new prefix pulls a different set of predictions. One caveat: Google’s autocomplete documentation says predictions take into account where a query comes from and what’s trending, so check from the country you actually serve.
2. People Also Ask and Related Searches
Search the seed and open three or four of the People Also Ask questions. The box grows every time you click, and the new questions tend to get narrower. Then scroll to the related searches at the bottom of the page.
I copy these into a plain list as I go. Nothing fancy, just a text file. PAA questions are already phrased like a reader’s question, which saves a lot of rewriting when they become H2s or FAQ items later.
3. Your Own Search Console Queries
This is my favorite source, and honestly the most underused. In the Performance report, add a Query filter, switch it to Custom (regex) and paste this pattern to show only queries of five words or more:
^(\S+\s+){4,}\S+$
It matches “how often should i feed a sourdough starter” and skips “sourdough starter”. I explain how that pattern works, plus nineteen others, in my guide to Google Search Console regex. Sort the result by impressions and you’re looking at long tails Google already connects with your site.
Two limits are worth knowing. Google says the interface exports up to 1,000 rows, and very rare queries are anonymized for privacy and never listed, according to its Performance data deep dive. So you see a large slice of your long tail, never all of it.
4. Forums, Reddit and Q&A Sites
Communities show you the words people use before any tool has counted them. Search Google for your topic plus “reddit”, or use site:reddit.com crate training, and read the thread titles. Titles are often near-perfect long tail keywords, written in the reader’s own language.
Read the replies too. Skimmers miss the gold. When the same follow-up question shows up in thread after thread, that’s a long tail worth a section, even if no tool lists it.
5. Your Site’s Internal Search
If your site has a search box, visitors are typing long tail keywords into it every day. In Google Analytics 4, the site search event in enhanced measurement fires when a results page loads with a query parameter, and by default it looks for q, s, search, query and keyword, per Google’s Analytics help. The terms land in the Search term dimension.
What I look for here is failure. A search term that people repeat but your site has no good answer for is a page you should probably write.
6. Paid Tool Filters
Every major keyword tool has some version of two filters: a word-count filter and a questions filter. Semrush’s Keyword Magic Tool has a Questions button and a word count option under Advanced Filters, and Ahrefs’ Matching terms report in Keywords Explorer has a Questions toggle and a word count filter. The pattern is the same anywhere: enter the seed, set word count to 4 or more, then switch on questions.
My take? Use the volume column to rank the list, not to cut it. Plenty of good long tails show 0 or a tiny number, and my post on zero volume keywords covers how to check whether real demand sits behind that zero before you write anything.
7. Your Sales and Support Conversations
Customer emails, sales calls and support tickets are full of long tail phrasing. “Can I use this on a rented apartment wall?” is a keyword. Nobody’s tool knows it exists yet.
I treat this as a cross-check rather than a primary source. When a phrase from a customer also shows up in autocomplete or Search Console, I move it to the top of the list.
A Worked Example From Seed to Page Plan
This example is illustrative. I made up the business and the phrases to show the method, and I’ve left out volumes because they’d differ in your market anyway.
Picture a dog trainer with a small site and a seed of “crate training”. After 30 minutes in autocomplete, PAA and two Reddit threads, the raw list holds about 40 phrases. Here’s a slice:
| Long tail phrase | Where it came from | Specific need |
|---|---|---|
| crate training a puppy at night | Autocomplete | Night routine |
| puppy cries in crate at night | Reddit thread title | Night routine |
| how long can a puppy stay in a crate | People Also Ask | Time limits |
| crate training schedule by age | Autocomplete | Time limits |
| crate training an older rescue dog | Reddit thread title | Adult dogs |
| what size crate for a labrador | Autocomplete | Crate size |
Six phrases. Four needs. That’s the pattern you’ll see on almost every list: far fewer distinct needs than phrases. The next section shows how I decide which needs get their own page.
How Do You Group Long Tail Keywords Into Pages?
Search each phrase and compare the top results. If two phrases bring up mostly the same ranking pages, Google treats them as one need, and one page should target both. If the results barely overlap, they need separate pages. My guide to SERP based keyword clustering explains the overlap threshold most people use.
Ahrefs has a useful pair of labels for this. It calls a phrase that’s a variation of a bigger query a supporting long tail, and a phrase that’s a distinct topic of its own a topical long tail. Supporting long tails become sections, H2s or FAQ items inside a bigger page. Topical long tails get their own URL.
If the top results for a phrase are the same pages that rank for your main keyword, then add it as a section to that page. If the results are different pages answering a narrower question, then it’s a topical long tail and earns its own post. If you can’t tell, then default to a section; you can always split it out later once Search Console shows the page getting impressions for that phrase.
In the dog trainer example, “crate training a puppy at night” and “puppy cries in crate at night” would most likely share results, so they become one page. “What size crate for a labrador” probably pulls up product guides and size charts, which suggests a separate page with a different format.
Branch 1: An Established Site With Search Console Data
Start with the regex filter, not with Google. Your data beats guesses. Your own long tail queries already have impressions, which means Google sees your site as relevant to them.
Sort by impressions and look at the landing page for each query. If a long tail lands on a page that only mentions it in passing, that’s an easy win: add a proper section that answers it. If a group of long tails lands on your homepage or a loosely related post, you’ve found a new page that should exist. My guide on Google Search Console keyword research shows the full workflow.
Branch 2: A New Site With No Data Yet
A new site has no queries to mine, so it works the other way round. Start with autocomplete, PAA and communities, then use a tool only to sort what you found.
In my experience, new sites should lean hardest on the long tail. A brand new domain has almost no authority, and long tail results pages are where weaker sites can still reach page one. I’d pick one tight group of related long tails, publish the pages for it, and wait for Search Console data before expanding. Your first impressions then tell you which direction Google already trusts you in.
Edge Cases That Change the Method
- Local businesses. Many local long tails carry no town name at all (“emergency plumber open sunday”). Check them from inside your service area, since autocomplete and results shift by location.
- Ecommerce stores. Product long tails (“waterproof hiking boots wide fit”) often belong on a filtered category page, not a blog post. Let the results page decide the format.
- Non-English markets. Autocomplete and forums work in any language, but translated keyword lists don’t. Collect long tails from native-language searches.
- Very new topics. If a topic is weeks old, no tool has data. Forums and autocomplete are your only real sources.
Decision Matrix: Which Method First?
| If your site | And you have | Then start with |
|---|---|---|
| Gets impressions in Search Console | No paid tool | The 5+ word regex, then autocomplete for gaps |
| Gets impressions in Search Console | A paid tool | The regex first, then tool filters to add volume |
| Is new, no data | No paid tool | Autocomplete, PAA, Reddit |
| Is new, no data | A paid tool | Tool question and word-count filters, then forums |
| Has a busy internal search | Any setup | Failed internal searches before anything else |
When Is It Worth Getting Help?
Finding long tails is easy. Choosing is hard. Deciding which 30 of 400 deserve pages, and which page each one belongs to, is where most people get stuck. If you’d rather hand that part off, my team covers it in our keyword research services, and you can try it on 5 keywords free before paying anything.
If you want to see what your competitors already rank for before you start, my companion guide on how to find competitor keywords walks through it.
Frequently Asked Questions
What Is an Example of a Long Tail Keyword?
“Best running shoes for flat feet on a treadmill” is a long tail keyword, while “running shoes” is a head term. The long tail version gets far fewer searches, but it tells you exactly what the searcher needs, which makes it easier to answer and usually easier to rank for.
How Many Words Is a Long Tail Keyword?
There’s no fixed number. Most long tails run three words or more, but the defining trait is low search volume and a specific need, not length. Ahrefs points out that some one-word keywords get very few searches, while some long phrases get hundreds of thousands.
Can I Find Long Tail Keywords for Free?
Yes. Google autocomplete, People Also Ask, related searches, Search Console, Reddit and your own site search cost nothing. Paid tools add volume estimates and speed, but the free sources often surface phrases the tools haven’t recorded yet.
Should Every Long Tail Keyword Get Its Own Page?
No. Most long tails are variations of a bigger query and belong as sections or FAQ items on one page. Only give a long tail its own URL when the results page for it shows different pages from your main keyword.
Last updated: October 2026 by Mizanur Rahman
