How to Do Keyword Research: A Field Guide in 7 Steps

What is keyword research?

Keyword research is the systematic process of uncovering and prioritizing the words and phrases your audience types into search engines, alongside the intent behind each query. The goal isn't to find the highest-volume word. The goal is to find the most relevant, most convertible and most reachable keyword for your page.

For us, the process never starts without understanding the client's product and business model. "Copy the competitor's keyword list" usually stalls within a few months, because every brand has a different value proposition and therefore a different user to catch.

Why does it matter so much?

The wrong keyword brings the wrong user. The wrong user clicks but doesn't convert. On one B2B SaaS client we drove 22k organic visits in the first 3 months from high-volume but poorly-matched keywords; trial sign-ups were near zero. After rebuilding the page set against the right intent, we shipped 8k visits but 140+ trials per month.

Intent: the signal that beats volume

Over the past few years Google has put intent ahead of keyword match. Two people typing the same word can want completely different things. Here are the four intent categories we tag every keyword with:

Fastest way to read intent: type the keyword into Google and look at the first 10 results. Whatever Google is ranking, that's the intent. We do this on every keyword we evaluate.

Types of keywords

Modern research groups keywords by tail length, not by volume:

• Head terms: 1-2 words, high volume, high competition (e.g. "SEO"). Out of reach without strong authority.
• Body terms: 3-4 words, mid volume, mid competition (e.g. "local SEO guide"). The healthiest starting point.
• Long tail: 4+ words, low volume but high intent (e.g. "local seo agency for bursa"). Faster to rank, higher conversion.

The 7-step modern process

1) Clarify the goal and persona

Who are we writing for? Decision-maker, end user, technical team? Without that clarity, the seed list is meaningless. We run a 30-min session with the client to define 3-5 personas.

2) Build the seed list

We extract 15-25 root words the personas would use. Gold mines: the words the client's sales team uses on calls, the language in customer support transcripts and the free-text answers in existing forms.

3) Expand with tooling

We expand the seed list using Google Search Console, Ahrefs, Semrush and Google Suggest. The output is usually 800-2000 keywords.

4) Tag intent

We label every keyword with one of the four intent categories. Skipping this step is the single biggest reason teams build the wrong pages next.

5) Cluster into topics

We group keywords that serve the same intent and are semantically related into a single "topic cluster." Each cluster gets exactly one target page. This prevents internal cannibalization and concentrates authority on one URL.

6) Prioritize

Which cluster goes first? Our math: (1) intent fit × (2) estimated conversion value ÷ (3) ranking difficulty. We pick the 8-12 clusters most likely to ship results in 3 months.

7) Monitor and refresh

Research is never one-and-done. Every 4-6 weeks we read fresh Search Console queries and check which clusters are growing. Markets shift, intent shifts, and research has to shift with them.

Our toolkit

Prioritization: where to start

The most common mistake we see: starting with the highest-volume keyword. New sites have almost no chance of ranking head terms in the first 6 months. Our quick 3-question test:

1. 1Can we produce the target page in 4-8 weeks?
2. 2Is the ranking difficulty (KD) compatible with our current authority? (We usually start under KD 30.)
3. 3Will ranking translate into a concrete business outcome (form, demo, sale)?

Anything that doesn't get three yeses goes into the backlog, not the first wave.

5 most common field mistakes

• Skipping intent and chasing volume. High volume can mean the wrong user.
• Building 4-5 different pages for the same intent and causing internal cannibalization.
• Stuffing every keyword onto the homepage. Each cluster needs its own page.
• Cloning a competitor's list and forgetting your own market position.
• Treating research as a one-off and working from the same list 6 months later.

Practical starter checklist

1. 1Define 3-5 personas and the language they actually use.
2. 2Pull 15-25 seed words from sales and support transcripts.
3. 3Expand the list to 800-2000 keywords with Search Console + Ahrefs/Semrush.
4. 4Tag every keyword with one of the four intent labels.
5. 5Group keywords into topic clusters; assign exactly one target page per cluster.
6. 6Use the 3-question test to pick 8-12 clusters for the first wave.
7. 7Refresh the list every 4-6 weeks using new Search Console queries.

Good keyword research is roughly 40% of an SEO program. The remaining 60% is content quality, technical health and authority. If you want us to build a full research and content plan for your brand, reach out via the contact page.