Keyword Research in the Age of AI Search

“What’s the best keyword?” is the wrong question, and it’s been the wrong question for a while now. Search has moved from matching exact phrases to understanding intent, which means chasing a single high-volume term is a much weaker strategy than understanding the range of ways real people actually ask for what a business offers. This guide walks through how keyword research actually works today, from understanding what someone means when they search, through to finding the right terms, using them properly, and adapting to how AI has changed search itself.

Start With Intent, Not the Word Itself

Before picking any keyword, it’s worth understanding what category of intent it represents, because the same word can mean completely different things depending on where someone is in their decision-making. Someone typing “running shoes” is usually still exploring — this is informational or broad commercial intent, and they’re unlikely to buy from the first page they land on. Someone typing “buy Nike Pegasus 41 women’s” has transactional intent — they know exactly what they want and are close to purchasing. A third type, navigational intent, is when someone searches a specific brand or website name because they already know where they want to go, such as searching “Matrix Internet” rather than “web design agency Dublin.”

Why This Distinction Changes Strategy

Targeting the wrong intent for a page is one of the most common keyword mistakes. A product page optimised for a broad, informational term will attract visitors who bounce immediately because they weren’t ready to buy, which can actually hurt rankings over time by signalling to search engines that the page isn’t a good match for that query. Matching content type to intent — blog content for informational searches, product and service pages for transactional ones — matters more than the raw search volume of any individual term.

Volume Without Intent Is a Trap

A keyword with enormous search volume but low relevance to what a business actually sells will bring visitors who leave immediately, which does nothing for rankings or revenue. A shop specialising in handmade wedding invitations doesn’t benefit from ranking for “invitations” broadly — that term pulls in people looking for birthday party invites, corporate event invites, and free templates, almost none of whom are a fit. Ranking well for “handmade wedding invitations Dublin” brings a smaller number of visitors, but a far higher share of them are actually looking for exactly that.

Long-Tail Keywords Still Do the Heavy Lifting

Longer, more specific phrases tend to convert better precisely because they signal a person further along in deciding what they want. Someone searching “best trail running shoes for wide feet” has a much clearer intent than someone searching “running shoes,” and a page built specifically to answer that narrower question will usually outperform a generic one competing against every major retailer for the broad term. This hasn’t really changed in years, but it’s still one of the most underused approaches for smaller sites trying to compete against much bigger ones.

How This Plays Out for Local Businesses

A business serving a specific area gets an additional layer of this for free: location-based long-tail terms. A car detailing business in Cork isn’t realistically going to outrank national chains for “car detailing,” but a page built around “car detailing services in Cork” — genuinely written for a Cork audience, not just keyword-stuffed — has a real shot at ranking well precisely because fewer sites are competing for that specific combination.

Finding the Right Keywords in Practice

Good keyword research starts with tools, but tools only tell part of the story. Google’s own Keyword Planner is free and gives a reasonable baseline for search volume, though it’s built primarily for advertisers and can be vague on exact numbers for smaller or more niche terms. Paid tools like Ahrefs and Semrush go considerably further, showing not just volume but ranking difficulty, the actual pages currently ranking for a term, and what related questions people are asking. AnswerThePublic is particularly useful for surfacing the informational, question-based searches a keyword tool alone might miss — the kind of long-tail phrasing real customers use rather than the tidy terms a marketer might guess at.

Learning From Competitors

Looking at which keywords competitors already rank for is often faster than starting from a blank page, particularly for a smaller business trying to find gaps larger competitors haven’t fully covered. A competitor gap analysis — comparing a business’s own ranking keywords against a competitor’s — tends to surface terms with genuine demand that haven’t been claimed yet, which are usually easier to rank for than the obvious, heavily contested ones everyone already targets.

Search Itself Has Changed the Question

AI-powered search tools now answer a large share of queries directly, often without the user ever clicking through to a website. This changes what “the best keyword” even means — visibility increasingly depends on being structured clearly enough, and being a trusted enough source, for those tools to pull from and cite. That still starts with genuinely understanding what real people are asking, just extended a step further than it used to be. Google’s own SEO Starter Guide covers the technical fundamentals this all sits on top of.

What Changes When AI Answers the Question Directly

The practical shift is in what “ranking well” actually gets a business. AI-generated summaries often answer a query directly at the top of the results page, and a growing share of searches end there without a click at all. Getting pulled into that summary depends less on matching a keyword and more on whether the content is structured clearly enough — direct answers stated early, genuine headings that map to real questions — for an AI system to extract and trust as a source. This tends to reward exactly the kind of specific, well-targeted content long-tail keyword research already produces, rather than requiring a separate strategy. A page built to genuinely answer “car detailing services in Cork” thoroughly is naturally well suited to being cited by an AI summary, in a way a thin, keyword-stuffed page never would be. Search Engine Land’s guide to AI search is a solid, regularly updated reference for how this continues to develop.

Relevance and Difficulty Still Matter Together

None of this means high-volume keywords should be ignored entirely — it means they need to be weighed against relevance and realistic difficulty, not chased for their own sake. A term with strong search volume, genuine relevance to what’s being offered, and a competitive landscape a smaller site can realistically break into is worth real investment. A term that only has one of those three qualities usually isn’t. We’ve written in more depth about the considerations that actually drive keyword success, which goes further into balancing these factors properly.

Putting Keywords to Work on the Page

Finding the right keyword is only half the job — where and how it’s used on the page matters just as much. A target keyword belongs naturally in the page title, the main heading, and early in the body content, but forcing it into every sentence reads poorly to a human visitor and can actually work against rankings, since search engines have gotten considerably better at penalising content that feels stuffed rather than written. Related terms and natural variations — synonyms, plurals, the phrasing a real customer would actually use in conversation — tend to help more than repeating the exact same phrase mechanically throughout a page.

Internal Linking Reinforces the Signal

Linking between related pages on the same site, using descriptive text rather than generic “click here” links, helps both visitors and search engines understand how a site’s content relates to a given topic. A car detailing business with separate pages for interior detailing, paint correction, and mobile services should link between them naturally wherever relevant, reinforcing to search engines that the site has genuine depth on the broader topic, not just one isolated page.

Tracking, Not Guessing

Keyword research isn’t a one-off task — rankings shift, competitors adjust, and search behaviour changes over time, particularly as AI-driven search continues to develop. Google Search Console, free for any site owner, shows which queries are actually driving impressions and clicks to a site, often revealing terms a business never deliberately targeted but happens to be ranking for anyway. Reviewing this regularly, rather than setting a keyword strategy once and leaving it, is what separates a strategy that keeps improving from one that quietly goes stale.

The Actual Answer

There isn’t a single best keyword, and there never really was — there’s a best keyword for a specific business, a specific audience, and a specific stage in someone’s search for what they need. Getting that mix right — matching intent to content, going specific rather than broad, weighing relevance against realistic difficulty, and structuring content clearly enough for both people and AI systems to make sense of it — matters far more than chasing whatever term happens to have the highest search volume this month. Our SEO services team can help work out what that mix actually looks like for a given site.