Almost everybody doing SEO for the first time starts the same way: they think of five words they would use to search for their own business and try to rank for them. Those words are, as a rule, the shortest, the broadest and the most expensive — and they’re fought over by sites with a ten-year head start.
Choosing keywords is the one decision in SEO you can’t fix with better writing. If the topic is wrong, a perfect text doesn’t help.
Three numbers that describe every keyword
| Number | What it means | The trap |
|---|---|---|
| Volume | how many times it’s searched a month | high volume ≠ good business |
| Difficulty | how hard it is to enter the top ten | an estimate, not a law |
| Intent | what the person wants when they type it | the most important, and no tool has it |
The tool gives you the first two. The third you have to read yourself — and it’s the one that decides.
Intent: four kinds of question
The same industry, four completely different queries:
- “what is engineered wood flooring” — learning. Won’t buy today, maybe never.
- “laminate or engineered wood” — choosing. Close to a decision.
- “wood flooring price per m2” — looking at money. Very close to buying.
- “floor fitting Croydon” — looking for a fitter. Buying.
If you have little time and few pages, start from the bottom up. First the keywords looking for a fitter and a price, then the ones choosing, and only at the end the ones learning.
Most people do the opposite, because the learning keywords have the biggest volume — and then they’re surprised to have traffic but no orders.
The process in five steps
1. Write ten words off the top of your head
No tools. How you would search for what you sell. That’s a seed, not a list.
2. Run them through research
One word yields several hundred real queries, with volume and a difficulty estimate. This is where you discover what you’d never have guessed — for example that “boiler replacement cost” is searched five times more than “boiler repair”.
3. Look at what people actually ask
Alongside the keywords themselves, pull out the questions: “how long does it take”, “can you”, “which is better”, “how much does it cost”. Those questions are ready-made subheadings, and they make the kind of text search likes to quote.
4. Open the search results and look at who’s there
This shows up in no number at all. Type in the keyword and look at the first five results:
- If they’re Wikipedia, forums and news portals, the person is learning — there’s no buying.
- If they’re shops and pages with prices, there’s buying.
- If all five are large sites with thousands of pages, your new page isn’t getting in there this year.
- If there are weak pages among them — thin text, a site with no contact details, an article from 2019 — you have a chance.
This step takes a minute per keyword and is worth more than any difficulty estimate.
5. Sort and take the top ten
For every keyword that made it, write down: volume, who’s above, how close to buying. Take the ten with the best ratio of closeness-to-buying and weakness-of-competition — not the ten with the biggest volume.
Why small keywords are often worth more
Take two keywords for the same joiner:
| “joinery” | “fitted kitchen Croydon price” | |
|---|---|---|
| Searches a month | 8,100 | 40 |
| Chance of entering the top 10 | small | large |
| Out of 100 visitors, asking for a quote | 1 | 12 |
| Realistic enquiries a month | 0 | 4–5 |
The keyword with 40 searches brings work, the one with 8,100 brings nothing — because you’ll never rank for it, and even if you did, the people typing it mostly aren’t buying a kitchen.
Ten such small keywords make a serious month. And one more thing: small keywords are what people dictate to their phone and ask AI assistants — all of search is moving that way.
Six mistakes that keep repeating
- One keyword, several pages. Three of your pages compete for the same keyword, so Google doesn’t know which to choose and ranks none of them.
- Keywords in the wrong language for the market you’re actually selling to.
- Ignoring how people write. People search with and without accents, in full words and abbreviations — check both variants.
- Stuffing keywords into the text. Hasn’t worked since 2012, and makes the text unreadable.
- Choosing by volume, without opening the search results.
- No review. A keyword list is refreshed twice a year; what was searched two years ago isn’t the same now.
In short
- Volume and difficulty are estimates; intent is what decides, and you read it yourself.
- Start with keywords looking for a price and a supplier, then work up towards the ones learning.
- Before you take a keyword — open the search results and look at who’s already there.
- Ten small, specific keywords bring more work than one big one.
- A keyword list isn’t permanent; refresh it twice a year.
If you want to see what’s actually being searched for in your industry instead of guessing, keyword and question research does exactly that job.