How to Do Keyword Research (the Modern Way) with AI

Keyword research has a reputation for being tedious, and done the old way it is: thousands of rows in a spreadsheet, sorted by volume, most of them useless. The modern version is simpler and sharper — generate from the right sources, judge each keyword on three axes, then cluster the survivors into a plan. Here's how.

Where good keywords actually come from

Strong keyword lists don't start with a database dump. They start with sources you can reason about:

1. Seed terms — the broad words that describe what you do. Everything branches from these.
2. Audience questions — what your customers actually ask, in their words. These become long-tail, high-intent articles.
3. Competitors — the terms ranking sites in your niche already cover (and the gaps they don't).
4. Search features — autocomplete, People Also Ask, and related searches reveal real demand straight from the SERP.

Step 1 — Generate widely from a seed

Start from one seed keyword and expand outward across the sources above. Aim for breadth first — you want the obvious head terms and the non-obvious long-tail questions in front of you at once. Don't filter yet; that comes next.

Step 2 — Read each keyword on three axes

Volume alone is a trap. A keyword is only worth writing if all three of these line up:

• Search volume — how many people search it. Bigger isn't always better; huge terms are usually unwinnable.
• Ranking difficulty — how hard it is to crack the first page. The sweet spot is decent volume at low difficulty.
• Search intent — what the searcher wants (learn, compare, buy). Match it, or you'll rank for traffic that never converts.

Step 3 — Group keywords by intent and topic

A flat keyword list is raw material, not a plan. The leverage is in clustering: grouping related keywords into the pillars and supporting articles they naturally form. *“Email deliverability,” “avoid spam folder,” “email authentication”* aren't three separate posts competing with each other — they're one cluster. Clustering also prevents keyword cannibalisation, where two of your own pages fight for the same term.

Step 4 — Turn keywords into articles

Every surviving keyword becomes one article with one job: rank for that term, for that intent. Assigning a single target keyword per article is what keeps your pages from cannibalising each other and gives each piece a clear brief. The research is finished when every keyword has a home.

A prompt you can copy

“Do keyword research around [seed keyword] for [audience]. Group the keywords into topic clusters, give each a search intent, and flag the ones with strong volume and low difficulty to target first.”

Paste into RibatAI's prompt bar

Common mistakes to avoid

• Chasing volume alone — high-volume head terms are usually unwinnable for smaller sites.
• Ignoring intent — the right keyword with the wrong intent brings the wrong readers.
• Leaving keywords as a flat list — cluster them, or you'll cannibalise your own pages.
• Skipping difficulty — without it you can't tell a quick win from a years-long slog.