Keyword Research for the AI Era: Beyond Volume and Difficulty

I want to show you something that broke my brain a little when I first saw it.

Take "WordPress hosting." In a keyword tool, it shows up as one row: search volume, difficulty score, CPC. One keyword, one target. That's how we used to think about this.

Now go to Perplexity and type "WordPress hosting." Watch what it generates. It doesn't return one answer — it returns a disambiguation. It asks, implicitly, what kind? Cheap shared hosting for a personal blog? Managed WooCommerce hosting for a store doing $50k/month? Enterprise-grade infrastructure with SLA guarantees? Three different problems, three different buyers, three different pieces of content you should be writing.

That's the shift. "WordPress hosting" isn't a keyword anymore. It's a category containing multiple intent clusters. And if you're targeting the category without specifying the intent, you're probably ranking for traffic that doesn't convert.

Why search volume is the wrong starting point

Tools like Ahrefs and SEMrush are useful, but they create a specific cognitive bias: you start optimizing for what's measurable instead of what's valuable. High-volume keywords are usually high-competition, high-ambiguity terms dominated by brands with far larger domain authorities than yours. The math rarely works out.

The move I made was to invert the research process. Instead of starting with volume and working backward to content, I start with problems and work forward to queries.

When a client describes a pain point to me in a sales call, I write it down verbatim. Not the polished version — the exact words they used. "We don't know if our WordPress updates are breaking things before clients notice" becomes a more useful research seed than "WordPress maintenance." The verbatim version is closer to how someone who has that exact problem would phrase a search.

Prompting AI for pain points, not keyword lists

The common mistake: asking AI tools to generate keyword lists. The output is usually generic and overlaps heavily with whatever everyone else is doing.

Better approach: ask for problems. Something like "what are the top technical concerns for a WooCommerce store owner who has grown past their current hosting infrastructure?" gets you a list of specific friction points — slow checkout, failed payment webhooks, database timeouts during sales. Each of those is a search query waiting to happen, and each one has commercial intent baked in.

Once you have the pain points, validate them in Ahrefs or SEMrush. You're using AI for problem discovery and traditional tools for demand confirmation. They do different things well.

The three intents worth caring about

Not all search intent converts equally, and where you focus should depend on what you're trying to achieve.

Informational queries — "what is WP Rocket," "how does Redis caching work" — are being absorbed by AI Overviews at an increasing rate. You can still rank for these, but the traffic is declining. The value here is authority signaling, not lead generation.

Commercial queries — "best WordPress hosting for WooCommerce 2026," "WP Rocket vs NitroPack" — still drive meaningful clicks because people doing comparison research want to see the detail and don't trust a one-paragraph AI answer. These are worth competing for.

Transactional queries — "hire WordPress developer," "WooCommerce speed optimization service" — are short, specific, and high-intent. These are the ones where even minimal traffic can produce real revenue. Competition is lower than you'd think because most content marketers are chasing volume, not intent.

Zero-volume isn't the same as zero-demand

This is the one that took me the longest to fully accept. Keyword tools show "0" search volume for queries that clients have asked me verbatim. The tools are sampling aggregate data — they can't capture the long tail at that level of specificity.

The rule I use now: if I've heard it in a client call, it's being searched. The person in that call learned about the problem from somewhere, discussed it with colleagues, and eventually tried to look it up. They probably couldn't find a clean answer, which is an opening.

Long-tail queries with "0" volume often convert at a higher rate than anything with 1,000+ monthly searches. The specificity is exactly the point — someone searching "WooCommerce checkout fails after Redis upgrade" has a very specific problem and is ready to trust whoever can solve it.

What to do with this

Pull your last five sales calls or client emails. What specific problems came up? Write them down in the client's own words. Run those phrases through Ahrefs — not to find high volume, but to find adjacent queries and confirm the topic area has some documented demand. Then write content that answers the specific problem, not the broad category.

Keyword research isn't a spreadsheet discipline. It's listening for what people are actually struggling with and deciding which of those problems you're best positioned to answer.

The keywords follow from that. They don't lead it.

---

Want a keyword strategy built around your actual audience and service area? Get in touch.