How to Do Keyword Research: A Step-by-Step Guide for Any Business

Ranking well and getting cited in an AI Overview used to be nearly the same achievement. Ahrefs’ own research found only 38% of AI Overview citations now come from pages that also rank in the top 10 organically, down from 76% seven months earlier. Doing keyword research properly, instead of guessing at what customers type into the box, still decides whether content gets found at all, even as the reward for finding it splits into two increasingly separate races.

Why Keyword Research Still Matters Despite AI

A search term’s monthly volume still tells a business roughly how many people are asking that exact question, whether they end up reading an AI overview, a chatbot reply or a classic blue link. That number never stopped mattering. The part that changed sits downstream of it, in what happens once someone’s looking at the answer itself.

Write content around what a business wants to say instead of what people are genuinely searching for, and the usual result is traffic that never moves. Nine times out of ten it’s because the content answered a question nobody was asking in the first place.

A business’s own internal language rarely matches what a customer types. A roofing company might describe its work internally as “domestic roofing solutions.” A homeowner with water coming through the ceiling searches “roof repair near me.” Missing that mismatch is where keyword research usually goes wrong, before it’s even properly started.

Search intent earns the visibility now. An AI system deciding what to cite reads for relevance and clarity in a way an old ranking algorithm was never built to, and a high-volume term with the wrong intent behind it satisfies neither.

Starting With Seed Keywords

Learning how to do keyword research starts with the seed keyword, the broad, obvious term a business would use to describe what it is trying to sell. “Accountant,” “wedding photographer,” “running shoes.” Nothing clever, nothing long-tail, just the plainest possible description of the thing itself.

Five or six seed keywords cover most businesses, pulled straight from the core services or products genuinely on offer. Type each one into Google. The “People also ask” box and the related searches sitting at the bottom of the page turn one seed term into a dozen more specific ones inside a few minutes of scrolling.

Competitors are a second source that should also be checked. Google’s own Keyword Planner can turn a competitor’s domain into keyword ideas, but it won’t show what that competitor actually ranks for. Seeing the real picture, the exact terms they’re already ranking for, some of which a business would never have thought to search for on its own, takes a paid tool with its own ranking index, such as Ahrefs.

Using Google Search Console for Existing Opportunity

Search Console already holds a list of every query bringing traffic to a site, ranked by impressions, clicks and average position. Most businesses only check it for problems. It’s just as useful for finding opportunity sitting in plain sight.

A query ranking at position eleven or twelve, just off the first page, with decent impression volume already, is usually the cheapest keyword opportunity available to any business. The content already ranks for it. Strengthening that one page, adding depth, fixing the title tag, updating it with fresher information, often moves it onto page one faster than building something new from scratch if that’s the correct place for that keyword to rank.

Filtering the Search Console query report by impressions without clicks reveals a different problem. Pages show up in search and never get chosen. That mismatch usually points to a weak title or meta description rather than a ranking problem, and it’s one of the fastest fixes available in the whole keyword research process.

Understanding Search Intent: The Four Types

Every search sits somewhere on a spectrum from “just looking” to “ready to buy.” Informational searches want an answer, not a sales pitch. “How does keyword research work” is asking to learn something, and a page that opens with a hard sell will lose the reader.

How to carry out keyword research for navigational searches is almost a separate skill entirely, since the searcher already knows exactly where they want to go. Someone typing a brand name plus “login” isn’t comparing options; they’re trying to get into an account they already have. Ranking for these matters mostly for protecting a brand’s own territory in the results.

Commercial investigation sits in the middle, comparing options before deciding. “Best keyword research tools UK” for example is a searcher narrowing down a shortlist rather than being ready to purchase. Transactional intent is the final stage, someone who’s decided and just needs the fastest path to buying, exactly what “buy,” “price,” or “near me” attached to a term usually signals.

Matching content to the wrong stage of that spectrum wastes both the content and the ranking. A landing page built for a transactional search, dropped onto an informational query, sends a confused visitor straight back to the results.

Reading the actual search results for a term settles this part for keyword research. A term that looks informational on paper sometimes returns a page full of product listings and comparison tools. Whatever the phrase itself suggests, Google’s search results are the real answer to what it means.

Long-Tail vs Head Terms: Where to Focus First?

“SEO agency” is a short-tail keyword. Short, broad, fought over by hundreds of agencies with years of authority already built up. Narrow it down to something like “SEO agency for doctors” and the field of realistic competitors will shrink hugely.

That specificity converts. Anyone typing five or six words into a search bar usually already knows roughly what they want, which is why long-tail terms tend to close at a noticeably higher rate than broad ones. A new website with no backlink history stands a real chance of ranking for a handful of these within months. The same site chasing a competitive head term is usually still waiting years later.

Start long-tail. Build authority and traffic from a cluster of specific pages ranking well first, then go after the head terms that cluster naturally feeds into once there’s something to compete on.

Keyword Difficulty vs Search Volume

A keyword difficulty score estimates how hard a term is to rank for, based mostly on the backlink profiles of whoever currently occupies page one. It’s an estimate, not a guarantee, and it varies noticeably between tools measuring the same term.

Chasing the highest search volume term available is one of the most common keyword research mistakes a growing business makes, made worse by ignoring difficulty entirely while doing it. Five hundred searches a month at low difficulty converts real traffic within weeks on a site with barely any backlinks yet. Five thousand searches at sky-high difficulty can sit unranked for years on that exact same site.

A simple grid fixes this, volume plotted on one axis and difficulty on the other. Low difficulty sitting together with decent volume, in one corner of that grid, is where a new or growing site puts its first few months of effort. High volume and high difficulty together in the opposite corner waits until there’s enough domain authority built up to genuinely compete for it.

The technical SEO and crawlability foundations of a site affect how quickly a term with moderate difficulty gets won, since a site Google struggles to crawl properly competes at a disadvantage regardless of how good the content sitting on it is.

Mapping Keywords to Pages

Every keyword needs exactly one home. Two pages both targeting the same term cause keyword cannibalisation, splitting the site’s own ranking signal between them instead of letting one page build it properly. The result? Neither page is likely to gain first page positions.

A keyword map handles this properly: one spreadsheet, one row per keyword. Each row records the page that owns the term, its intent, and its current ranking. Reviewing it every quarter catches pages that have quietly drifted into competing with each other as new content gets published around them.

Pillar and cluster pages give that map an actual structure to organise around. A pillar page targets the broad head term, “keyword research,” say, covering the topic at a level wide enough to link out to everything underneath it. A handful of supporting cluster pages then own the specific long-tail terms sitting underneath that pillar, each one built around a narrower slice of the topic and linking back up to the pillar page directly.

An SEO keyword strategy built around pillar and cluster pages keeps the keyword map honest. Every cluster page has a clear, single keyword attributed to it in the map, and the pillar page collects the authority all of them build over time by linking to and from each one. A content team publishing without this structure usually ends up with several posts all circling the same broad phrase, none clearly attributed to it in any map, and none benefiting from the others’ existence at all.

Link building to support keyword rankings targets pillar pages specifically once this structure exists, since a link pointing at a page with one clear keyword attributed to it in the map passes considerably more useful signal than one aimed at a page competing for several terms at once.

Free vs Paid Tools: What You Actually Need

Google’s own Keyword Planner, built for advertisers but usable by anyone with a free Google Ads account, remains the single best free source of genuine search volume data. Google Search Console and Google Trends round out a genuinely usable free stack, enough to run every step already covered here without spending anything.

Semrush’s keyword research tools and Ahrefs’ Keywords Explorer are where most keyword research tools UK agencies end up paying once free tools stop covering the scale needed. Both pull from billions of tracked keywords, cluster related terms automatically, and score difficulty across an entire competitor’s site in one pass rather than one term at a time. Semrush tends to edge ahead on advertising and PPC-adjacent data; Ahrefs is usually the sharper pick for backlink-heavy difficulty scoring specifically.

A small business publishing a handful of pages a month rarely needs a paid service. An agency managing several client accounts, or a business tracking hundreds of terms across a large site, usually does, since manually checking difficulty and volume one term at a time stops being realistic well before that scale is reached. The keyword research process at that scale genuinely does need the automation a paid tool provides.

Learning how to do keyword research properly with the free stack first, before paying for either tool, builds a genuine understanding of what the underlying data means. Someone who’s never manually checked a “People also ask” box or read through a page of actual search results tends to trust a difficulty score blindly. A score built from a huge international backlink index can read far tougher than it actually is for a niche B2B term with only a handful of real competitors.

Frequently Asked Questions

What is keyword research?

Finding the exact words and phrases people type into search engines, then using that data to decide what to write. Evidence instead of guesswork.

Why is keyword research important?

It decides whether content ever gets found. A well-written page still sits unread if nobody searches for the phrase it was built around.

What’s the best keyword research tool?

Google’s own Keyword Planner covers most of what a small business needs for free. Semrush and Ahrefs are the strongest paid options once a business needs difficulty scoring and competitor analysis at scale.

How do you do keyword research for SEO specifically?

Start with seed keywords for the core services or products, expand with Search Console and competitor research, then map each term to a single page.

If you need help getting it right, our London SEO agency can take it from here.