Keyword research for beginners: start with what your competition ignores
The first client I ever landed came from a keyword with 40 searches a month. Forty. My competitor had 12,000 monthly visitors and never touched that phrase, because his tool showed it in grey. I built one page around it, and that page still brings me roughly 15 qualified leads a year — people who arrive with a problem I can solve, not a browsing habit.
That's the whole game of keyword research for beginners: not finding the biggest term, but finding the one nobody else bothered to write about. Most guides hand you a definition, a "find, analyze, use" diagram, and send you off. You don't need a diagram. You need a way to decide which phrase deserves your next Saturday afternoon.
Here's what I actually do, including the part where I wasted two months chasing a keyword that was never going to convert.
Key Takeaways
- Start with seed phrases taken from real conversations, not from a tool's suggestion box.
- Volume matters less than the ratio between intent and competition.
- Modifiers ("for beginners", "free", "how to") change the audience more than they change the topic.
- Skip a keyword if you cannot imagine the person who typed it.
- Google Keyword Planner and Semrush are useful — but only after you know what you're looking for.
- Re-check a keyword every six months. AI answers have quietly killed some click-heavy terms.
What keyword research actually is (and what it isn't)
Keyword research is the practice of finding out which words people type into a search box when they have a problem, then deciding whether you can answer that problem better than the pages already ranking.
The definition is the easy part. The hard part is that most beginners treat the tool's output as an answer, when it's only an inventory. A list of 800 keywords is not a strategy. It's a spreadsheet with anxiety in it.
The question behind the query
Every phrase carries a hidden situation. "Keyword research tool" is someone comparing software. "How do I find keywords nobody is targeting" is someone already frustrated with the obvious answers. Same topic, completely different article.
When you read a keyword, ask what the person had already tried before typing it. That single habit separates useful research from busywork, and it costs nothing.
Why beginners should ignore volume at first
New sites cannot win on crowded terms. If your domain is three months old and you target "keyword research", you're competing against tools that have published on that phrase since the 2010s. You will lose, and you'll conclude that SEO doesn't work.
Long-tail phrases are not a consolation prize. They're the only place a new site can build authority fast enough to matter. I'd rather rank first for eight small phrases than sit on page four for one big one.
Step 1: build a seed list from real life, not from a tool
Open a blank document. Before any software, write down ten questions your audience has asked you out loud. Client emails, support tickets, comments on someone else's post, the weird thing your colleague asked at lunch.
Those are your seeds, and they're more accurate than any suggestion engine, because suggestion engines recycle what's already published. If it appears in the dropdown, someone has already covered it.
Where to find phrases nobody lists
- Forum threads — the exact wording people use when they're stuck
- Your own inbox, sorted by "how do I…"
- YouTube comments under tutorials on your topic
- Sales call notes, if you have any
- The footnotes of articles that rank on page two (they ran out of space, you didn't)
Notice these sources are slow and manual. That's the point. The list you build in an hour of reading beats the export you download in ten seconds, because you understand every entry on it.
Where Google Keyword Planner fits
Google Keyword Planner is free with an Ads account, and it's the obvious starting point. Use it to check whether a phrase has any search activity at all, and to see how Google groups related terms.
What it won't tell you is whether the searcher is ready to act. Its ranges are wide, its intent data is thin, and it doesn't show you what the results page looks like. Treat it as a spell-checker for demand, not a decision-maker. I've seen it flag a phrase as "low volume" that went on to bring me 60 visits a month, because the number it shows is a bucket, not a measurement.
Step 2: read the results page before you write a single word
This is the step beginners skip, and it's where most of the value sits. Before you commit, search the phrase yourself and look at what Google decided to show.
If the first screen is all product pages and shopping results, that's a commercial query and your tutorial will struggle. If it's forum threads and Reddit discussions, Google thinks people want opinions, not guides. If an AI-generated answer fills the top of the page, fewer people will click anything at all.
A comparison that helps you decide
| What you see in results | What it means | Worth targeting? |
|---|---|---|
| Forums, Reddit, Quora | Searchers want lived experience | Yes, if you have a story |
| Product and category pages | Buying intent | Only with a commercial page |
| Thin listicles from big sites | Nobody has gone deep | Strong yes |
| AI answer above the links | The question is fully answered on the page | Usually no |
| Mixed guides and videos | Format matters as much as topic | Yes, if you can match the format |
I once spent six weeks on a guide for a phrase where the entire first page was official documentation. I ranked ninth, got almost no clicks, and learned that you cannot out-authority a manufacturer on its own product terms.
Step 3: pick metrics that match your situation
Three numbers matter, and only three: search demand, the strength of the pages already ranking, and how well the phrase matches what you sell or explain.
Using Semrush (and that competition score)
Semrush keyword research gives you volume, a difficulty score, and a cost-per-click estimate. The difficulty score is the useful one, and the cost-per-click tells you something subtler: a high CPC means advertisers are paying real money for that phrase, which usually means commercial intent.
For a beginner site, my rough thresholds are simple. Under a difficulty score of about 30 and you can compete within a few months. Between 30 and 60, you'll need genuinely better content and a few internal links. Above 60, park it until your site has history.
When the numbers disagree, trust your judgment
Tools contradict each other constantly. One shows 200 searches, another shows 1,900. That's not a malfunction — they sample different data. When two tools disagree by a factor of five, both are guessing, and you should weight the phrase on how well you can answer it instead.
Frankly, I've stopped opening three tools per keyword. One for demand, one for competition, and my own eyes on the results page. That's it.
Step 4: prioritize, publish, then measure
A list of 200 keywords is useless until you sequence it. I sort mine into three piles: publish now, publish later, and never — and the "never" pile is usually the biggest.
- Publish now: low competition, clear intent, you can write it from experience
- Publish later: good phrase, needs a page you haven't built yet
- Never: high difficulty, vague intent, or nothing new you could add
The measurement part is where most beginners lose momentum. Give each page eight to twelve weeks before you judge it. If it's getting impressions but no clicks, your title needs work. If it's ranking on page two and stuck, it needs internal links and a more specific angle. If it gets zero impressions after three months, the phrase doesn't have the demand the tool promised.
I keep a simple sheet: phrase, publish date, current position, clicks this month. Twenty rows, checked monthly. It took me a year to realize that's all the reporting I actually needed.
How keyword research has changed, and what to do about it
Something shifted recently that most beginner guides haven't caught up with. Search engines now answer a lot of questions directly on the results page, and a growing share of younger users start their product research inside short-video apps rather than a search bar.
The practical consequence: some phrases you'd have targeted a few years ago now produce zero clicks even from position one. You won't see this in a volume column. You'll only see it in your own numbers, months later.
So I now weight phrases differently. Questions with a single factual answer are riskier. Questions where people want an opinion, a comparison, or a story are safer, because a summarised answer can't replace them. "What is a canonical tag" is fragile. "Why my canonical tags kept breaking" is not.
And if your audience lives in video, keyword research still applies — you're just researching titles and captions instead of headings. The words are the same. The surface isn't.
The mistakes I made so you don't have to
Early on I targeted a phrase with good volume and a difficulty score I liked, wrote 3,000 words, and got 11 visits in three months. The reason was embarrassingly simple: I'd never searched the phrase myself. The results page was full of job listings. Nobody typing it wanted an article.
Another time I built an entire cluster around a topic I found mildly interesting but had no experience with. The writing was fine. It read like fine writing by someone who had nothing to say.
Two rules came out of those failures. Search the phrase before you commit to it, and never target something you can't speak about from memory. If you have to research your own opinion, pick a different keyword.
The beginner advantage isn't knowledge — it's that you haven't yet learned which phrases are "supposed" to be worth chasing. Keep that. Pick the small, awkward, overlooked question, answer it better than the ten pages above you, and let the volume grow while everyone else fights over the same crowded terms.
