Keyword Research for Beginners: A Repeatable 7-Step Process

Almost every failed blog post we are asked to rescue has the same root cause: it was written before anybody checked what people actually search for. Keyword research is the step that prevents that, and it is far less mysterious than the tool vendors would like you to believe.

This is the process we run on every client project and teach inside the SEO Starter Pack. It takes an afternoon the first time and about an hour once you know it. You do not need paid software to complete it, although paid software makes step four faster.

Why keyword research decides everything downstream

Search engines answer questions. If nobody is asking the question your page answers, no amount of optimisation will produce traffic. That sounds obvious written down, yet the most common content mistake we see is a page built around how a business describes itself rather than how customers describe their problem.

A real example from a client project: a firm of accountants had a beautifully written page called "Statutory compliance services". Nobody searches for that. The same service, described as "annual accounts for limited companies", had 1,300 monthly searches in the UK alone. Same service, same page, one word change in the title and heading — and within eleven weeks that page was the firm's second biggest source of enquiries.

Keyword research is also what protects you from writing content you cannot win. Every market has terms that are effectively closed to a new site: "web design", "personal injury lawyer", "car insurance". You can spend a year writing brilliant content for those terms and never crack page three. Research tells you that before you start, not after.

The three questions every keyword has to pass

1. Does anybody search for it? If the answer is genuinely zero, the page can still exist — it just should not be your priority.
2. Do the people searching it want what I sell? "How to do SEO yourself" and "SEO agency near me" have wildly different commercial value.
3. Can I realistically rank for it within twelve months? Judged against who currently ranks, not against the tool's difficulty score in isolation.

Step 1: Build a seed list from your own sales language

Do not open a keyword tool yet. Tools expand lists; they cannot invent the starting point, and their suggestions are biased toward whatever is already popular. Your seed list should come from four places you already own:

• Your enquiry forms and inbox. Search your email for the last thirty enquiries and copy the exact phrases people used to describe what they wanted.
• Sales call notes. The question a prospect asks twice on every call is almost always a keyword.
• Your services and products, written plainly. Not the clever internal name — the plain-English version a stranger would use.
• Google Search Console. Open Performance → Queries and sort by impressions. These are terms Google already thinks you are relevant for, which is enormously useful signal.

Aim for thirty to sixty seed phrases. Messy is fine. Duplicates are fine. You are collecting raw material.

Step 2: Expand each seed into the long tail

Now bring in the tools. Put each seed into two or three of the following and export everything:

Two things matter more than which tool you pick. First, always capture the modifiers: "near me", "cost", "price", "best", "vs", "for small business", "in Chester". Modifiers are where intent lives. Second, keep the long tail. A term with 40 searches a month and obvious buying intent is usually worth more to a small business than one with 4,000 searches and none.

Expect to end this step with several hundred rows. That is normal and it is about to get much smaller.

Step 3: Label every term by search intent

Intent is the single most useful column in your spreadsheet and the one most beginners skip. Label each term with one of four values:

• Informational — "how does seo work", "what is a title tag". The person wants to learn. Blog content.
• Commercial — "best seo course", "seo agency vs freelancer". The person is comparing. Comparison pages, reviews, case studies.
• Transactional — "buy seo course", "seo audit price". The person is ready. Service and product pages.
• Navigational — "seo starter pack login". They want a specific place. Usually already handled.

How do you tell? Search the term and look at what Google is already rewarding. If the first page is all blog posts, Google has decided the intent is informational, and a service page will not rank there no matter how well written it is. If the first page is all product pages, the reverse applies.

The results page is not your competition. It is Google showing you, for free, exactly what format it believes answers this query. Argue with it and you lose.

This is also how you catch the mismatch that quietly kills conversion: ranking a sales page for an informational query brings visitors who bounce, and ranking a blog post for a transactional query brings visitors who cannot buy.

Step 4: Judge difficulty honestly

Every paid tool gives you a difficulty score out of 100. Treat it as a first filter, not an answer, because those scores are calculated mostly from backlinks and cannot see relevance, freshness or intent match.

The manual check takes ninety seconds per term and is far more reliable:

1. Search the term in an incognito window.
2. Look at the ten organic results. How many are national brands, Wikipedia, Amazon or government sites? If it is most of them, move on.
3. How many are businesses roughly your size? If two or three are, the term is winnable.
4. Read the top result properly. Could you produce something genuinely more useful — more specific, more current, better illustrated, actually written by someone who does the work? If yes, you have a shot.

Step 5: Cluster terms into pages

This is the step that turns a keyword list into a plan. Group every term that could be answered by the same page into one cluster. "seo checklist", "on page seo checklist", "seo checklist for blog posts" and "seo checklist pdf" are one page, not four.

The test is simple: if two terms show substantially the same top-ten results, they belong on the same page. If the results differ, they need separate pages.

Give each cluster:

• A primary term — the one you will use in the title tag and H1.
• Supporting terms — used naturally in subheadings and body copy.
• A target URL — either an existing page or "new".
• An intent label, inherited from the terms in it.

Clustering also prevents keyword cannibalisation, which is one of the most common problems we find in audits: three or four thin pages competing with each other for one query, none of them strong enough to rank. Consolidating those into one page is frequently the single highest-impact change on a small site.

Step 6: Map clusters to your site architecture

Open your sitemap — or just list your pages — and put each cluster against a URL. You will find three cases:

Work top-down: money pages first (your services), then the commercial comparison content, then the informational blog content that feeds them. It is tempting to start with the blog because it feels easier. Resist. Optimising the page that sells the thing usually pays back within weeks, while a blog post takes months to mature.

Step 7: Prioritise and put it in a calendar

A plan you never action is a spreadsheet. Score every cluster on two axes — potential value and effort — and sequence them.

We use a simple 1–5 scale for each and sort by value divided by effort. The top of that list is almost always: existing service pages with impressions but poor positions. Those are the pages where twenty minutes of work moves a page from position 14 to position 7, and position 7 gets clicks where 14 gets none.

Then commit the order to actual dates. Two pages a month, done properly, beats ten pages in a burst followed by six months of nothing — both for rankings and for your own sanity.

Six mistakes that waste the most time

1. Chasing volume. Ten visitors who need your service beat a thousand who are curious.
2. Ignoring intent. The most common cause of "we rank but nobody enquires".
3. One page per keyword. Creates thin pages that cannibalise each other.
4. Trusting difficulty scores alone. Always look at the actual results page.
5. Never revisiting. Redo the research annually; search language changes faster than you think.
6. Researching, then writing whatever you fancy. The brief has to carry the cluster through to the draft, or the research was decoration.

Frequently asked questions

How long does keyword research take?

The first full pass on a small business site takes four to six hours spread over a couple of sittings. Annual refreshes take about an hour. Per-page checks before writing take ten minutes.

Do I need paid tools?

No, but they save time. Search Console, autocomplete, People Also Ask and Keyword Planner will get you a solid plan for free. Paid tools mainly speed up difficulty scoring and competitor gap analysis.

How many keywords should one page target?

One cluster, which typically means one primary term plus five to twenty close variants. If you find yourself wanting to cover two clearly different questions, that is two pages.

What about AI search and chat answers?

The underlying job is unchanged: be the clearest, best-sourced answer to a specific question. Pages that do that well are the ones being cited in AI answers, and the research process for finding those questions is exactly the one above.

Lyndsay Cambridge

Content & on-page lead

Lyndsay turns keyword research into content people actually want to read. She writes most of the training inside the SEO Starter Pack.

More about the studio