How to Carry Out Keyword Research: The Complete Agency Guide
Table of Contents
Keyword research is the process of finding the words and phrases your customers actually type into Google before they buy from you, and using that data to decide what to build, write, and rank for. Get it right, and a website attracts visitors who convert. Get it wrong, and you spend money on pages nobody finds.
For businesses across Northern Ireland, Ireland, and the UK, that distinction matters more than it once did. Search has split into two surfaces: traditional blue links and AI-generated answers in Google AI Overviews, ChatGPT, and Perplexity. A keyword strategy built only for the old rankings misses half the picture. This guide covers both, with a section dedicated to what changes for UK and Irish businesses specifically.
“Too many businesses guess at what their customers want,” says Ciaran Connolly, Director at ProfileTree. “Keyword research removes the guesswork. It tells you exactly what problems people are trying to solve and gives you a roadmap for creating content that genuinely helps them.”
What Is Keyword Research?
A keyword is the specific word or phrase someone types into a search engine when looking for information, a product, or a service. Keywords fall into a few useful categories.
Short-tail keywords contain one or two words, such as “SEO” or “web design.” They attract high search volumes but face intense competition and rarely signal clear intent on their own.
Long-tail keywords run to three or more words, such as “SEO services for small businesses Belfast.” They attract fewer searches individually but deliver more qualified traffic from people closer to a decision.
Branded keywords include a company name, for example, “ProfileTree digital marketing.” Commercial intent keywords signal readiness to act, such as “hire,” “price,” or “quote.”
This is the foundation of any keyword strategy. Everything that follows in this guide builds on getting these categories right before a single page gets written.
Why Keyword Research Is Changing
Search behaviour has moved on from matching strings to matching meaning. Google and the large language models behind ChatGPT and Perplexity now interpret what a searcher wants, not just which words they used. A page built to “rank for a keyword” in the old sense can still miss the underlying question entirely.
Two shifts matter most for SME owners planning a keyword research and strategy project this year.
First, AI Overviews and chat-based answers now appear above or instead of traditional results for an increasing share of informational queries. A business that only tracks blue-link rankings is no longer seeing the full picture of its search visibility.
Second, the keywords worth targeting have changed shape. A single high-volume head term used to be the prize. Today, a cluster of related questions, answered clearly and linked together, often earns more total visibility than chasing one big keyword in isolation.
This does not replace the fundamentals of keyword research techniques covered later in this guide. It changes what you do with the keywords once you have found them. Businesses that want a structured way to translate Search Console data into a working keyword strategy without managing it in-house can read more on ProfileTree’s search engine optimisation services page.
The Keyword Research Workflow
Professional keyword research strategies combine several approaches into one process. Each step reveals something the others miss.
Step 1: Understand Your Audience Before You Open a Tool
Effective keyword research starts with understanding who you are trying to reach and what they actually need, not with software.
Speak to your current customers about how they found you and what language they used to describe the problem you solved. Businesses often discover they describe themselves one way while customers search a completely different way. A business might call itself a “digital transformation consultancy” while its prospective customers search “help implementing AI in a small business.”
Study what competitors rank for and which of their pages attract engagement, without copying them outright. Look instead for the questions they are not answering. Online communities such as industry forums, LinkedIn groups, and local Facebook groups also reveal the casual phrasing people actually use, which often differs from formal industry terms.
Step 2: Build Seed Keywords and Topic Clusters
Seed keywords are the core terms that describe what a business offers. List 10 to 15 of them: your services, your location, the problems you solve, and the alternative names customers might use.
Group seed keywords into topic clusters. For a digital agency, clusters might include web design and development, SEO services, content marketing, and AI solutions. Each cluster becomes a pillar of content rather than a single isolated page, which also improves how a site is structured for both users and search engines. Businesses planning a new site around these clusters often find it more efficient to involve a web design team at this stage, since the keyword map should shape the site’s navigation and page list, not get bolted on afterwards.
Step 3: Use Tools to Expand and Validate
Google Keyword Planner provides UK-specific search volume directly from the source, useful for seasonal trends. The competition score it shows reflects paid advertising competition, not organic ranking difficulty.
SEMrush and Ahrefs add keyword difficulty scores, cost-per-click data, and “people also ask” questions. Answerthepublic and QuestionDB focus specifically on question-based searches, which are useful for FAQ planning and for the featured snippets that often sit above standard results. Ubersuggest and the Keyword Surfer browser extension offer a lower-cost way to quickly validate ideas.
No tool replaces judgment. A keyword difficulty score tells you how hard a term has been to rank for historically; it does not tell you whether you can create something better than what is currently ranking.
Step 4: Decode Search Intent
Search intent falls into four categories, and matching content type to intent matters more than matching the exact keyword.
Informational intent: the person wants to learn something, such as “what is keyword research.” Navigational intent: the person wants a specific site, such as “ProfileTree blog.” Commercial investigation: the person is comparing options, often signalled by “best,” “vs,” or “review.” Transactional intent: the person is ready to act, using words like “hire,” “price,” or “quote.”
Search for the keyword yourself and see what currently ranks. If the results are all blog posts, the intent is informational, and a service page will not rank there. If the results are company sites with clear pricing, the intent is transactional.
Step 5: Prioritise by Difficulty and Business Value
Not every keyword deserves equal attention. Quick-win keywords have a difficulty score below 30, often with monthly volumes of 10-100 searches and clear commercial intent. Strategic growth keywords sit in the 30 to 50 difficulty range, achievable within six to twelve months with better content and some accumulated authority. Long-term authority keywords sit above 50, the broad industry terms central to a business that take longer but are worth pursuing for topical authority, even before they rank.
| Difficulty Score | What It Means | Realistic Timeline |
|---|---|---|
| 0 to 30 | Achievable for newer or smaller sites | Weeks to a few months |
| 31 to 50 | Needs decent content and some authority | Six to twelve months |
| 51 to 70 | Needs strong domain authority and excellent content | Twelve months or more |
| 71 to 100 | Typically dominated by major brands | Long-term authority play, not a quick win |
Keyword Research for Northern Ireland, Ireland, and UK Businesses
Almost every major keyword research guide online is written for a US audience, using US spelling and US search volumes. That gap is an opportunity for businesses operating in Belfast, Dublin, or anywhere else across these markets, because the search behaviour, the spelling, and even the currency in cost-per-click data differ.
UK searchers use UK spelling. “Optimisation” and “optimization” can carry different volumes in keyword tools, and a tool defaulting to US English will under-report the UK-specific term. The same applies to “centre” versus “center” and to local terminology such as “solicitor” rather than “attorney,” which represent the same role but different search behaviour entirely.
Add geographic modifiers to core service terms: service plus location (“web design Belfast”), location plus service (“Belfast digital marketing agency”), and wider regional terms beyond the obvious city name, such as County Antrim, Ulster, or specific neighbourhoods. Businesses serving customers across Northern Ireland and the Republic of Ireland should also account for cross-border search behaviour rather than assuming one keyword list covers both markets.
A Belfast-based accounting firm chasing “tax planning Northern Ireland” will usually find more qualified traffic there than competing for the generic, nationally dominated term “accountant.” The same principle applies to a video production company: people search “how much does a company video cost” far more naturally than “corporate videography pricing,” and content built around that exact phrasing tends to outperform content written in formal industry language. Businesses that uncover this kind of search behaviour through their own keyword research and then need to act on it, by commissioning the video itself, can see examples of the format on ProfileTree’s video production page.
For businesses without an in-house marketing function, this kind of localisation work is also where a Belfast-based or Northern Ireland-based local SEO team has a structural advantage over an agency operating from elsewhere: the search behaviour around specific towns, cross-border terms, and regional spelling differences is not abstract knowledge but daily, local context.
Optimising Keyword Research for AI Overviews and Generative Search
Google’s AI Overviews and AI tools such as ChatGPT, Perplexity, and Gemini now generate summarised answers ahead of, or instead of, the traditional list of results for a meaningful share of informational queries. This changes which keywords are worth targeting and how the content built around them needs to be structured.
AI Overviews tend to draw from pages that already rank in the top 20 of traditional organic results, and they favour sections written as direct, self-contained answers. Practically, this means a “what is keyword research” section needs a clear, concise definition near the top rather than buried after three paragraphs of preamble, and a step-by-step process benefits from genuine numbered steps rather than narrative prose, since that structure is easier for an AI system to extract and cite.
A second, less obvious shift is in which keywords still send a human to your website at all. A purely informational query such as “what is keyword research” might be fully answered inside an AI Overview, sending no click at all. A more specific, judgement-dependent query, such as “how to implement AI in a Northern Ireland retail business” or “SEO strategy for a professional services firm,” still requires the kind of personalised expertise an AI summary cannot fully replace. Targeting these more specific, advice-led queries is often a better use of content effort than chasing high-volume generic terms that AI systems are increasingly answering directly.
Businesses that want to build AI literacy into their own marketing function, rather than treating this as something only an agency can interpret, can find more detail on ProfileTree’s AI training and implementation page, or the specific application of AI inside SEO and content work on the AI-enhanced marketing page.
Mapping Keywords to Content and Pages
Gathering keyword data only creates value once it shapes what gets built. Different keywords need different page types, matched to the intent identified earlier.
Service pages should target transactional keywords with clear commercial intent: “web design services Belfast,” “SEO consultant Northern Ireland,” or “AI training for businesses UK.” These pages need to be specific, action-oriented, and clear about who they are for.
Blog content should target informational and commercial investigation keywords: “how to improve website loading speed” or “what to include in a web design brief.” This is where keyword research marketing efforts compound over time, since useful blog content continues attracting search traffic long after publication.
Pillar pages target competitive head terms with full, in-depth coverage, supported by detailed blog posts that target the long-tail variations underneath them.
A practical example: an SME that completes its own keyword research and discovers strong, consistent search volume for a service it does not currently have a dedicated page for, commonly “ecommerce,” “website maintenance,” or a specific industry niche, has effectively written its own brief for a new page. Where that gap points to a full site restructure rather than a single new page, that is a website development conversation rather than a content one, since the technical build needs to support the keyword map from the start.
Choosing a Keyword Research Agency in the UK
Not every business has the time or the in-house resource to run this process properly, and that is a reasonable thing to outsource rather than a failure. The biggest mistake businesses make when evaluating a keyword research agency in the UK is judging the pitch rather than the process.
Ask any agency three direct questions before committing. What does your process actually look like step by step, from seed keywords through to a finished content brief? Will you show your working, including which tools you used and why, rather than just handing over a spreadsheet of terms? How do you decide which keywords to prioritise, and on what basis: search volume alone, or volume weighed against difficulty and genuine commercial value to the business?
A keyword strategy report that lists hundreds of terms with no prioritisation and no connection to actual page plans is not a strategy. It is raw data. The value sits in the steps that follow: clustering, intent mapping, content planning, and ongoing tracking through Google Search Console. Businesses that want this handled end to end, from the initial research through to published, ranking content, can see how that process works on ProfileTree’s content marketing services page.
Monitoring Performance Over Time
Keyword research is not a one-off task. Use Google Search Console to monitor which queries bring traffic, the average position for each one, and how click-through rates shift over time. Set up rank tracking through a tool such as SEMrush, Ahrefs, or Moz, and review monthly rather than waiting for an annual audit.
If a keyword generates impressions but no clicks, or clicks but high bounce rates, the content is not matching what the searcher actually wanted, regardless of where it ranks. Revisit the page against the original search intent before assuming the keyword itself was wrong.
When a page already ranks for several related terms without reaching the top of page one for any of them, expanding that page’s coverage of the topic is often more effective than starting a new page from scratch.
FAQs
How often should a business carry out keyword research?
Run a full keyword research project annually, with a lighter quarterly review to catch new trends or competitor movement. Monitor existing keyword performance monthly through Google Search Console rather than waiting for the annual review to spot a problem. Launching a new service or entering a new market is a trigger for focused research outside this usual cycle.
Which keyword research method works best?
No single method covers everything. Audience research and seed keywords give direction, tools provide volume and difficulty data, competitor analysis reveals gaps, and search intent analysis confirms whether the content type matches what the searcher wants. Combining all four produces a far more complete picture than relying on tool output alone.
How do I judge keyword difficulty for my own site?
Tools such as SEMrush, Ahrefs, or Moz provide a 0 to 100 difficulty score, but a manual check matters just as much. Search the term, look at who currently ranks, and judge their domain authority and content quality honestly against your own site. A score of 0 to 30 is generally achievable for a newer site; 70 and above usually requires established authority that takes time to build.
What is the biggest mistake businesses make in keyword research?
Relying on tool output without checking intent and context. A tool can tell you a term gets 500 searches a month; it cannot tell you whether those searchers want to buy, browse, or simply learn something. Search your target terms yourself before committing a content plan to them, and avoid chasing high-volume keywords that have no real connection to what your business actually offers, since that traffic will not convert.
Does keyword research still matter with AI Overviews and ChatGPT now part of search?
Yes, and arguably more than before. AI systems still draw heavily from pages that already rank well in traditional search, so the underlying keyword and intent work has not become optional. What has changed is the need to structure content so it can be extracted and cited clearly, alongside ranking for it in the conventional sense.
How do I do keyword research for a brand-new website with no existing data?
Without historical Search Console data to lean on, focus on long-tail, lower-competition terms with clear, specific intent rather than competing immediately for broad head terms. Seed keywords drawn from direct conversations with early customers, combined with a manual SERP check for each candidate term, will get a new site further than tool data alone in the first few months.
Taking the Next Step
Keyword research only creates value once it is applied consistently to what gets built, written, and tracked. Start with one priority topic cluster aligned to a core service, build content around both the head term and its long-tail variations, and set up monthly tracking before moving on to the next cluster. The businesses that see the clearest results from this process are rarely the ones with the most sophisticated tools. They are the ones that actually act on what the data tells them, month after month.