Why AI works surprisingly well for keyword research

Traditional keyword research tools — Semrush, Ahrefs, Moz — are built around search volume data. They tell you how many times a keyword was searched last month. Useful. But they don’t tell you why someone is searching, what they actually want, or what they’ll do next.

AI does something different. Because it’s been trained on enormous amounts of text — including search queries, forum discussions, product reviews, help articles, and customer feedback — it has a deep intuitive grasp of how people talk about problems. That’s exactly what good keyword research requires.

Ask ChatGPT “what would someone search for when looking for a plumber in Cape Town who fixes geysers?” and it will give you 20 variants you’d never have thought of — including the long-tail questions real customers ask before they’re ready to pick up the phone.

What AI can’t do — be honest about this

AI tools don’t have real-time search volume data. ChatGPT doesn’t know that “solar geyser installer Cape Town” gets 260 searches a month in South Africa. It can’t tell you which keywords your competitors are ranking for right now. And it can’t show you what’s changed in the last six months.

What this means in practice: use AI for generation and intent analysis, then validate the shortlist with a real data source — even just Google’s autocomplete and People Also Ask, or your own Google Search Console data if you have a site already.

The 5-step AI keyword research workflow

Copy-ready prompts

Replace anything in [square brackets] with your specifics. Use in ChatGPT, Claude, or Gemini.

I run a [type of business] in [city/region], South Africa. My target customer is [describe customer]. I help them with [describe problem you solve].

Generate 30 keywords my potential customers would search on Google. Split them into three groups:

GROUP 1 — Commercial intent (ready to hire/buy): keywords showing they want to take action now
GROUP 2 — Informational intent (researching): keywords showing they're learning before deciding
GROUP 3 — Local intent (location-specific): city and suburb-level variants

For each keyword, add one sentence explaining the search intent — what is this person actually trying to do?

Format as a table: Keyword | Group | Intent explanation

I am a [your business type] based in [location], South Africa.

List 25 questions a potential customer would type into Google before hiring someone like me. Include:
- Questions about cost and pricing
- Questions about how to choose / what to look for
- Questions about the process / what happens
- Questions about common problems or concerns
- "Is it worth it" type questions

Format each as a question exactly as someone would type it into Google. Then add: which of these has the highest commercial intent (closest to a buying decision)?

I have this list of keywords for my [business type] business in South Africa:
[paste your keyword list here]

For each keyword:
1. Flag any that use terminology more common in the US/UK that South Africans might search differently
2. Suggest South African alternatives or variants where relevant
3. Identify any keywords where SA-specific context matters (e.g. SARS, load shedding, local regulations, rand pricing, specific SA platforms)
4. Add suburb or city variants for the top 5 local-intent keywords

I'm based in [your city] and serve [your area/region].

My business is [describe your business] in [location], South Africa.

My main competitors are: [list 2-3 competitor names or website URLs]

Based on what you know about businesses like these, what topics and keywords are they likely targeting that I might be missing? Specifically:

1. Service pages they probably have that I don't
2. Location pages they probably target
3. Informational content that builds trust before a sale
4. Long-tail question keywords in our industry

Give me 15 keyword/content gap opportunities, ranked by how likely they are to drive qualified leads.

SA-specific keyword research tips

Search in Afrikaans too

Depending on your market, Afrikaans keywords can be completely uncontested. A plumber in the Western Cape might find “loodgieter Kaapstad” has almost no competition. Ask the AI to generate Afrikaans variants if your customer base includes Afrikaans speakers.

Suburb-level local keywords

In South African cities, people search by suburb more than city. “Accountant Sandton” vs “accountant Johannesburg”. “Dentist Claremont” vs “dentist Cape Town”. Generate suburb-level variants for your top 5 service keywords — especially if you serve a specific area.

SA-specific problem keywords

Load shedding, SARS submissions, POPIA compliance, BEE certification, UIF claims — these are problems unique to South Africa that generate search traffic with almost no competition. If any of your services touch these issues, build pages around them.

Price-sensitive searches

South African searchers are highly price-conscious. “How much does X cost in South Africa” keywords convert extremely well. Include “cost” and “price” variants in every commercial keyword set you build.

How to validate what AI gives you — free options

FAQ