How to find low competition keywords (and stop wasting months on dead ends)
Most people search for low competition keywords the wrong way around. They open a tool, sort by volume, look at the first page, pick something that sounds nice, and then spend three months writing articles nobody finds. I know, because I did exactly that for the first year of my site.
Then I flipped the order. I stopped looking for keywords and started looking for gaps. The difference is enormous, and it has nothing to do with which tool you pay for.
This isn't a listicle of tools. It's the method I actually use now, the numbers I check, and the traps that cost me roughly 40 articles before I figured out what was going on.
Key Takeaways
- Low competition isn't a number in a tool. It's a situation on a search results page: weak domains, thin content, unanswered questions.
- Sort by volume and you'll pick fights you can't win. Sort by weakness and you'll find openings that tools rate as "hard."
- Three free checks beat one paid metric:
allintitle:search, SERP domain authority eyeballing, and whether the page actually answers the query. - Keyword difficulty scores are directional, not verdicts. They miss local results, forums, and featured snippets.
- The best low competition keywords are usually questions nobody wrote a proper answer to yet.
- Realistic threshold for a new site: competing pages with fewer than 15 referring domains and no exact-match title.
Why your keyword tool keeps sending you into walls
Here's a scene I remember clearly. A site I was working on had a difficulty score of 12. Twelve. Green light. I wrote a 2,200-word article, published it, waited. Six months later it sat on page four, behind three Reddit threads, a Quora answer, and a YouTube video.
So what happened? The tool measured backlinks to the exact phrase. It didn't measure that the query was clearly informational and that Google had decided discussions were the best answer. No amount of link building fixes that.
The score is a proxy, not a diagnosis
Every difficulty metric I've used—Ahrefs' KD, Semrush's Keyword Difficulty, Moz's PA—does roughly the same thing: it looks at how strong the pages currently ranking are, then compresses that into one number. Useful as a filter. Dangerous as a decision.
What the number can't see:
- Whether the top results are forums, videos, or product pages rather than articles
- Local intent, which changes everything about who you're competing with
- Whether a featured snippet already swallows the click
- How thin the actual content is (a 400-word page with high authority still loses to a good 1,800-word answer)
I'll be blunt: if you pick keywords based only on a green score, you're outsourcing your judgment to a formula that has never seen your site.
The threshold that actually matters
Forget the score for a second. When I open a results page now, I ask one question: could a page on my site, with my current authority, plausibly beat these specific results?
My working benchmark, after a lot of trial and error: if more than two of the top five results have 15+ referring domains pointing at them, I walk away. If most of them are under 10 domains and none has the exact phrase in the title, that's a genuine opening.
The manual method: three free checks in ten minutes
No subscription needed for this part. You need a browser and a bit of patience.
Check 1 — the allintitle: search
Type this into Google:
allintitle: "your keyword phrase"
This returns pages that use the exact phrase in their title tag. Why does it matter? Because a page with the exact phrase in the title is a page that was deliberately built for that query. Those are your real competitors.
If you get 40 results, good. If you get 4,000, that phrase is saturated and you should look at a longer variant. I've found phrases where the count was under 20 and the top results were blog posts from 2019 that had never been updated. Those are gifts.
Check 2 — read the results page like a detective
Not skim. Read. For each of the top five results, note three things:
- Domain type (blog, forum, shop, news site)
- Whether the title contains the query
- Whether the page genuinely answers the question, or just mentions the topic
An example from my own notes: a phrase with a difficulty score of 34. Four of the top five results were e-commerce category pages that mentioned the term once in passing. One was a forum thread with a single useful reply. I published a proper 1,600-word guide and hit position three in about seven weeks. The score said no. The page said yes.
Check 3 — mine the questions and the forums
Scroll to the "related questions" block on any results page and write down every question. Then search Reddit and Quora for the same phrase plus "how" or "why." Threads with lots of upvotes but no definitive answer are the single richest source of low competition keywords I know of.
Why? Because those threads prove demand exists, and the absence of a good article proves supply doesn't.
If you're paying for Ahrefs or Semrush, use them differently
I've paid for both. They're worth it, but not for the reason most people buy them.
Stop using the keyword explorer as a starting point. Use it as a verification step after you've already found a candidate manually. Here's my workflow:
- Find the phrase through SERP reading, forums, or your own analytics
- Check its referring-domain profile for the top three results
- Check whether the phrase has a parent topic you can also cover
- Only then look at the difficulty score, and treat it as one input among five
Which tool for which job
| Tool | Best used for | Weakness I've run into |
|---|---|---|
| Google Keyword Planner | Broad volume ranges, free, useful for validating demand | Volume buckets are wide; no competition insight at all |
| Google Trends | Checking whether interest is rising, flat, or dying | Relative numbers only, useless for absolute volume |
| Ahrefs | Referring domains per result, parent topic mapping | KD ignores intent mismatch and forum results |
| Semrush | Question-based keyword lists, competitor gap analysis | Difficulty scores skew high on niche topics |
| Manual SERP reading | Everything that actually decides whether you can rank | Slow, doesn't scale past a few dozen checks a week |
That last row is the honest one. Manual work doesn't scale, and that's precisely why it still works.
Finding low competition keywords on YouTube and Reddit
Both platforms have their own logic, and neither respects the metrics you're used to.
YouTube: judge the videos, not the search volume
On YouTube, the competition is visual and much easier to read. Search your phrase, then look at the top results and ask:
- Are the top videos more than two years old?
- Do they have under 20,000 views?
- Are the thumbnails low effort, or the titles vague?
- Does the video actually answer the query in the first 30 seconds, or does it spend two minutes on an intro?
A phrase where the top three videos are old, low-view, and poorly titled is wide open. I've seen channels rank a brand-new video in the top five within days on phrases like that. Search volume figures on YouTube are notoriously unreliable, so lean on the competition signals instead.
Reddit: read the unanswered questions
Search a subreddit for your topic and sort by "top" for the past year. Then look for posts with strong engagement and no satisfying answer in the comments. Those posts are keyword research handed to you for free, with proof of demand attached.
One caution, and it's a real one: don't just rewrite a Reddit thread. Answer the question better than the thread did, with structure and specifics. Otherwise you're competing with the thread on its home turf, which you'll lose.
Questions I get asked constantly
Can you find low competition keywords for free?
Yes, and honestly the free route is where I'd start even with a budget. The three checks above cost nothing but time. Google Keyword Planner and Google Trends cover demand and direction. The manual SERP read covers competition, which is the part paid tools get wrong most often. The only thing you give up is speed at scale.
Should I ever target high search volume keywords?
Not first, unless your site already has authority. High volume almost always means high competition, and a new site ranking on page four gets no traffic at all. I'd rather own ten phrases with 300 monthly searches each than fight for one with 20,000 and never appear. Build the small wins, then use that authority to climb.
How many keywords should I target at once?
Fewer than you think. I cap it at five active targets per site at any time, because each one needs a genuinely better page than what's ranking. Ten half-finished articles lose to one thorough one, every time.
What actually moves the needle
After all the tools and checks, the pattern is boring and consistent: the pages that rank are the ones that answer a specific question more completely than anything else on the results page.
Keyword research doesn't find those pages for you. It just tells you which questions are still unanswered. The rest—the depth, the specifics, the willingness to write 1,800 words where everyone else wrote 400—is on you.
And that's the part no difficulty score will ever measure. Which is exactly why it still works.
