How to Find the Right Keywords for Your Content

Most keyword research produces a spreadsheet nobody uses. It has thousands of rows, sorted by search volume, and every term at the top is one the site has no chance of ranking for. The list is technically correct and completely unusable.

Useful keyword research answers a much narrower question: which specific queries can this site realistically win, that are worth winning? That is a judgement call, but it is one you can make systematically.

Start from the problem, not the tool

Before opening any keyword tool, write down twenty things your customers actually say. Not marketing language — the words in support tickets, sales calls and the first message someone sends you.

This matters because tools only expand on the seeds you give them. Seed them with your internal vocabulary and you get a list about your internal vocabulary. Seed them with customer language and the whole list changes.

Classify intent before anything else

A keyword is not a topic — it is a request. Four intents cover almost everything, and mismatching them is the most expensive error in the whole process.

• Informational — 'what is technical seo'. Wants an explanation. Serve a guide.
• Commercial — 'best seo tools for agencies'. Wants to compare options before deciding. Serve a comparison.
• Transactional — 'hire seo consultant'. Ready to act. Serve a service or product page.
• Navigational — 'search console login'. Wants a specific destination. Usually not worth targeting unless it is yours.

You do not decide the intent — the search results do. Search the term and look at what ranks. If the first page is nine tutorials, that query is informational, and pointing a product page at it will fail no matter how well written the page is.

Judge difficulty by reading, not by the score

Every tool gives a difficulty number, and every tool computes it differently from an incomplete link index. Treat it as a rough sort, then do the real check by hand — it takes two minutes per term.

1. 01Search the query and look at the ten results. Are they household names in your market, or ordinary sites like yours? If it is all household names, skip it for now.
2. 02Are any results genuinely weak — thin, outdated, off-intent, or a forum thread? A weak result on page one is an opening.
3. 03Is the intent something you can serve better than anyone there? If you have first-hand experience the current results clearly lack, that is your edge.
4. 04Is there a path from the query to something you sell? Not a direct one necessarily, but a plausible one.

If you cannot articulate why your page would deserve to outrank the current number three, you are not ready to write it.

Volume is the least useful column

Search volume is an estimate, it is averaged over twelve months, and it says nothing about whether the traffic is worth having. A term with 200 searches a month from people ready to buy is worth more than 20,000 searches from people writing a school essay.

Longer, more specific queries are usually the better trade early on: less competition, clearer intent, and a much higher chance that the person searching wants what you offer. A hundred pages ranking for specific terms beats one page not ranking for a broad one.

Group into clusters, then map to pages

Search engines match meaning, not strings. Dozens of phrasings resolve to the same underlying question and should all be served by one strong page — not by one thin page each.

The practical test: search two keywords and compare the results. If the same pages rank for both, they are one page's job. If the results are different, they are two pages.

One page, not four — the SERPs for these overlap almost entirely:

  what is technical seo
  technical seo meaning
  technical seo explained
  technical seo basics

Separate pages — different SERPs, different intent:

  technical seo checklist       -> a checklist, probably with a download
  technical seo audit cost      -> a pricing / service page
  technical seo tools           -> a comparison

Assign each cluster exactly one target page. Two pages aimed at the same cluster will compete with each other, split their internal links, and both underperform the single page you should have written.

Prioritise with something you can defend

Once clusters are mapped, order them by a simple product rather than by volume: how likely you are to rank, times how much a visitor from that query is worth, times how often it is searched. Anything you cannot win goes on a later list, not a discarded one.

• Win now — low competition, clear intent, direct commercial relevance. Write these first.
• Build toward — valuable but competitive. These need supporting content and authority before they are attemptable.
• Not yours — high volume, wrong intent, or dominated by sites operating on a different scale. Note them and move on.

Then check whether you were right

Keyword research is a hypothesis. Search Console tells you what actually happened, and the most valuable report is the one nobody opens: queries where you rank between eight and twenty. Those are pages the engine already considers relevant but not quite best — usually the cheapest wins available.

Look at them monthly. Often the fix is not a new page but a paragraph the existing page is missing, which you can find by reading the two results currently above you.