How to find search terms that bring the right buyers

Good search engine keyword research is not about finding the biggest numbers. It is about finding the words buyers use when they are close to acting, then building pages that answer that need clearly. For a small business, that usually means fewer vanity visits and more of the right enquiries.

If your site is in German, Polish, Swedish, Lithuanian or any other non-English language, this matters even more. Direct translations often miss how people really search in that market. We see this often. A business has pages for the terms its team uses internally, while customers use different wording altogether. The result is traffic that does not convert, or no traffic at all.

What search engine keyword research is really for

The practical purpose of search engine keyword research is simple. We use it to match our pages to real demand, in the language buyers actually use, at the stage where they are ready to compare, enquire or buy.

That is different from chasing traffic.

A term can have high search volume and still be poor for business. If we sell payroll software for small firms in Ireland, ranking for a broad term like “payslip” may bring students, employees, and people looking for templates. Ranking for “payroll software for small business Ireland” is likely to bring fewer visits, but more relevant ones.

For most small and medium businesses, the best keywords sit in one of these groups:

• Service or product terms, such as accountant for contractors, CNC machining services, or fleet tracking software
• Problem-led terms, such as how to reduce warehouse picking errors
• Comparison or evaluation terms, such as best CRM for estate agents
• Local terms, such as tax adviser in Cork or physiotherapist in Leeds
• Brand plus service terms, where people already know us and want details

Each group reflects a different buying stage. Someone searching a broad educational query may be months away from buying. Someone searching for a service in a specific town may be ready now.

That is why keyword research should feed actual page decisions. Which services deserve their own pages? Which locations justify dedicated landing pages? Which recurring customer questions should become articles? If the research does not change what we publish, it is only a spreadsheet exercise.

For a small business site, we usually care about three outcomes:

• More qualified organic visits
• More enquiries, calls, demo requests or sales
• Better visibility in AI answers from tools such as ChatGPT, Gemini, Perplexity, Claude and Grok

Those outcomes often come from the same source. Clear pages, in the right language, covering the right topics with enough depth and specificity.

If you want a fuller view of how this works on a small business site, we covered that in what keyword research means for a small business site.

How to find the terms your customers actually use

The best keyword ideas usually come from customers first, tools second.

Start with the language already appearing in your business. Look at:

• Sales call notes
• Contact form submissions
• Email enquiries
• Live chat transcripts
• Support tickets
• Quotes and proposal requests
• Product reviews
• CRM records

Pull out repeated nouns, verbs and phrases. Pay attention to the exact wording customers use for their problem, not just your solution. A buyer may search for “late invoice reminders” rather than “accounts receivable automation”. A manufacturer may search for “food safe plastic parts” rather than a formal material name.

Next, speak to the people who hear customers every day. Ask sales and support:

• What do prospects ask for in their first message?
• What wrong terms do they use?
• Which words signal a serious buyer?
• Which requests usually lead nowhere?

Those answers help separate useful demand from noise.

Then use Google itself. Type the core term and note:

• Autocomplete suggestions
• “People also ask”
• Related searches at the bottom of results
• The wording used in top-ranking page titles

This is one of the quickest ways to expand a seed list. It also shows whether searchers want a guide, a product page, a local provider, or a comparison.

Use your own site data as well. In Search Console, open the Performance report and review:

• Queries with impressions but low clicks
• Queries ranking in positions 5 to 20
• Queries reaching the wrong page
• Country and device splits
• Brand versus non-brand searches

These are often your easiest wins. If a page already gets impressions for a relevant phrase, Google has some confidence in it. Improving that page can be faster than starting from nothing. We wrote more on this in Search Console impressions but no clicks - what to fix first.

If you do not have paid tools, that is not a blocker. You can still build a strong first list from customer language, Google suggestions and Search Console. We set out a practical process in how to find useful search terms without paying for tools.

For non-English sites, one extra rule matters. Do not translate an English keyword list and assume it will work locally.

That fails for three reasons:

• Buyers may use a different structure or word order
• The common term may be less formal than the dictionary translation
• Search intent may differ by country

German is a good example. Compound nouns can produce several valid forms, but only one may be the common commercial search. In the Nordics, English loanwords may compete with local terms, and the winner can vary by industry. In Poland and the Baltics, direct translation from English often sounds like internal company language rather than buyer language.

That is why at Seonis we research in the site’s own language, not from an English master list. Our system also runs every article through a language gate before publishing, because keyword relevance is lost quickly if the wording sounds translated.

Which keywords are worth targeting first?

Once you have a list, the next job is prioritisation. Not every relevant term deserves a page straight away.

We judge keywords on five practical factors.

1. Intent

Intent is the first filter. Ask what the searcher is trying to do.

Broadly, terms fall into these buckets:

• Informational, they want to learn
• Commercial, they want to compare options
• Transactional, they want to buy or enquire
• Navigational, they want a specific company or site

For a small business, commercial and transactional terms usually come first. Informational terms matter too, but mainly when they support a buying journey.

For example:

• “How does heat pump servicing work” is informational
• “Heat pump servicing cost Glasgow” is commercial
• “Book heat pump service Glasgow” is transactional

All three can be useful, but they belong in a sensible order.

2. Relevance

A term must match what we actually sell.

This sounds obvious, but it is where many keyword lists go wrong. If we offer implementation for a software product, a keyword about free templates may bring the wrong audience. If we supply premium joinery, ranking for bargain-led terms may create low-quality leads and pricing friction from the start.

A useful test is this. If someone searched this term, landed on our page, and read it carefully, would they be pleased with what we offer? If not, the keyword is not a fit.

3. Competition

Some terms are dominated by large directories, marketplaces, government sites or national publishers. That does not mean we ignore them, but we should be realistic.

If the first page is full of major brands with deep content libraries and thousands of links, a small business may get better returns from narrower terms first, such as:

• More specific service variations
• Industry-specific use cases
• Location-based searches
• Problem-led searches with clear intent

This is where long-tail terms earn their keep. They may have lower volume individually, but they are often easier to rank for and closer to conversion.

4. Location

For many businesses, geography changes everything.

A solicitor in Manchester should not build the same keyword plan as an e-commerce site shipping across the EU. A local clinic needs town and region terms. A B2B exporter may need country-specific pages in several languages. An agency serving Ireland may need separate treatment for searches in Dublin and Cork.

Where rules differ, say so on the page. The UK is outside the EU, so terms linked to VAT, consumer rights, customs, data handling or regulated services may need separate content from an EU-wide page. If you serve both, do not blur them into one generic article.

5. Business value

Finally, ask what a ranking is worth.

A keyword that brings ten serious enquiries a year may be more valuable than one that brings a thousand casual visits. We look at:

• Average order value
• Margin
• Sales cycle length
• Close rate
• Repeat business potential
• Whether the service is strategically important

This is where owners usually make better decisions than tools. Software can estimate search demand. It cannot tell you that one service line is far more profitable, or that your team is trying to fill capacity in a specific region.

How to group keywords into pages and articles

A keyword list becomes useful only when we turn it into a page plan.

The biggest mistake here is creating one page per tiny variation. That leads to thin content, internal competition and confusion about which page should rank.

A better method is to group keywords by shared intent.

If several terms would be satisfied by the same page, they belong together. For example:

• “accountant for contractors”
• “contractor accountant”
• “limited company accountant for contractors”

These can usually sit on one strong service page.

Use service pages for commercial and transactional terms. Use product pages for specific offers, categories or features. Use articles for informational and problem-led searches that support the sale.

A simple structure looks like this:

Service pages

These target core buying terms. They should cover:

• What we offer
• Who it is for
• Locations served, where relevant
• Process
• Pricing approach, if suitable
• Proof, examples or case studies
• Clear next step

Product or category pages

These suit e-commerce and software sites. They should focus on the exact item, category or use case people search for.

Articles

These answer questions, explain options, compare approaches and bring in earlier-stage searchers. Good article topics often come from objections in the sales process.

For example, if buyers often ask whether one material is suitable for outdoor use, that is an article. If they ask whether you serve a certain city, that may be a location page instead.

To avoid overlap, assign one primary topic to each page. Then support it with related secondary phrases that fit naturally. Keep a simple keyword map in a spreadsheet:

• Target page URL
• Primary keyword theme
• Secondary themes
• Search intent
• Country or language
• Status, planned, draft, live, update needed

This matters even more on multilingual sites. A page in Finnish should not be a direct clone of the Swedish page if search behaviour differs. The offer may be the same, but the keyword grouping may not be.

At Seonis, this is one reason we do not just generate translated articles in bulk. We write and publish natively, to fit the actual keyword clusters in that language. We also publish directly to WordPress, Shopify, Webflow and Ghost, because a plan only matters if it reaches the live site consistently.

Common keyword research mistakes small businesses make

The first mistake is chasing volume.

High-volume terms look attractive in a report. They often disappoint in the pipeline. We would rather rank for a specific query that produces enquiries than a broad term that attracts the wrong audience.

The second is translating keywords directly from English.

This is especially common in EU businesses with export markets. A team builds the strategy in English, then localises the wording later. That often produces pages that are technically correct but commercially weak. Search habits are local. Research needs to be local too.

The third is targeting terms that do not match the offer.

Examples are everywhere:

• A premium service targeting bargain-led searches
• A local firm writing national pages with no delivery capability
• A B2B supplier chasing consumer traffic
• A consultancy targeting software download terms

These mismatches create bounce, poor enquiries and wasted content effort.

The fourth is splitting hairs between near-identical terms.

If two phrases lead to the same result page and need the same answer, they usually belong on one page. Creating three weak pages where one strong page would do is a common self-inflicted problem.

The fifth is publishing generic AI content.

Readers can spot it. So can your sales team when the enquiries are poor. Generic content tends to have three flaws:

• It says what anyone in any country could say
• It misses the way buyers phrase the problem locally
• It does not reflect the actual service, process or constraints of the business

Used badly, AI makes keyword research look successful on paper while weakening trust on the page. Used properly, it can speed up drafting, clustering and routine updates, but it still needs native language control and commercial judgement.

That is why we built Seonis to work in the site’s own language, check content through a language gate, and report back in the owner’s language. The system runs every night, but it is still bounded by the evidence on the site and in the market. It cannot turn a weak offer into a strong one, and it cannot fix a page that promises something the business does not deliver.

How to measure whether your keyword choices are working

We do not judge keyword choices by rankings alone.

The first place to check is Search Console. Watch:

• Impressions for target pages
• Clicks and click-through rate
• Average position
• The actual queries each page is appearing for
• Changes by country, device and date

Early progress often looks like this:

• A page starts appearing for more relevant queries
• Impressions rise before clicks do
• Position moves from page three to page two, then page one
• Click-through improves after title and description changes

That is normal. Not every gain is immediate.

In Google Analytics or your analytics platform, check what those visits do next:

• Contact form submissions
• Calls
• Demo requests
• Purchases
• Quote requests
• Time on key pages
• Assisted conversions

If rankings improve but leads do not, revisit intent and page fit. You may be attracting the wrong audience, or the page may not be strong enough to convert.

In Google, also search the target terms manually from the right country and language settings. Look at the result types. Are product pages winning? Local packs? Guides? Video? If your page format does not match the result pattern, that is a clue.

For businesses that care about AI visibility as well as search traffic, measurement needs to go beyond rankings. We track whether ChatGPT, Gemini, Perplexity, Claude and Grok mention the business when buyers ask relevant questions. That does not replace search metrics, but it adds a useful layer. More buyers now ask AI tools for provider recommendations before they click through to a site.

Ongoing research matters because search behaviour shifts. New product names appear. Markets adopt new wording. Competitors publish into gaps. A keyword list from last year is often stale.

This is where software helps, if it is built for the language and market you actually serve. Seonis runs research, publishing and tracking overnight, in the website’s own language. It also builds authority through a members’ link exchange priced in credits, and reports results back in the owner’s language. For agencies and owner-managed firms without an in-house SEO team, that removes a lot of routine work while keeping the output checkable.

The main thing to measure, though, is still simple. Are we becoming easier to find for the searches that lead to business?

If the answer is yes, your keyword choices are working. If not, do not add more content blindly. Recheck the search terms, the intent behind them, and whether the page truly matches what the buyer wanted to find.