Keyword research

In the B2B SME sector, thorough research determines whether a blog post, landing page or product page will deliver qualified traffic for months to come or disappear into the index as an invisible file.

Contrary to popular belief, keyword research is not about compiling the longest possible list of popular words. It’s about evaluating search terms based on relevance, search volume, competition and, above all, search intent, and using this to establish a prioritisation that aligns with your business objectives. A term with 10,000 searches per month is worthless if searchers are looking for something other than what you offer. Similarly, a seemingly minor term with 80 searches per month could be the most profitable on your entire website if it precisely matches the query a decision-maker uses to contact your sales team.

Why keyword research is essential

Search engines connect demand with supply. Every search query is a documented need, and keyword research makes these needs visible and measurable. Four reasons make it an essential step:

• Customers’ language rather than the company’s. Internal product names and technical jargon often differ significantly from what customers are searching for. A manufacturer might refer to “warehouse management software”, whilst the target audience searches for “merchandise management system” or “warehouse management tool”. Only research can bridge this gap.
• Prioritising scarce resources. Content production costs time and money. Thorough research shows which topics are in demand and which you can leave out.
• Realistic expectations. The ‘Keyword Difficulty’ metric allows you to identify early on whether a term is even achievable or whether established competitors have firmly secured the top positions.
• Structure for the entire website. Research doesn’t just yield individual terms, but clusters of topics from which a sensible site architecture—comprising pillar pages and subpages—can be derived.

Added to this is a structural shift: according to a Sistrix analysis from February 2026, covering over 100 million German keywords, AI-generated answer boxes – known as ‘AI Overviews’ – now appear in around 20 per cent of all search queries. As a result, the click-through rate for the number one organic result has fallen from 27 per cent to 11 per cent – a drop of around 60 per cent. In total, around 265 million clicks are lost each month in Germany because 78.6 per cent of AI Overviews appear above the organic results. Keyword research remains central nonetheless, but its focus is shifting: away from pure volume, towards terms with genuine purchase intent and thematic depth, which also exist in zero-click search and for AI visibility. Anyone wishing to appear as a source in AI Overviews needs content that provides precise answers to well-researched questions.

Key metrics at a glance

Keyword research relies on a handful of metrics that you need to understand in order to evaluate terms effectively. The following table summarises the key metrics.

The keyword research process

A robust research process follows a repeatable procedure. The following phases have proven their worth in practice and can be applied to any project, from a single blog article to a complete website architecture.

1. Collect seed keywords

You start with seed keywords – that is, initial terms that fundamentally describe your business. For a Shopware service provider, these might include ‘create an online shop’, ‘Shopware agency’ or ‘e-commerce consultancy’. You can derive these seed keywords from your service portfolio, from discussions with the sales team, from support tickets and from the language customers actually use in their enquiries. The closer the seed keywords are to the language customers actually use, the better the subsequent expansion will be.

2. Expand keywords and enrich data

Tools generate hundreds of variations from the seed keywords. The Google Keyword Planner provides related terms and search volumes, but requires a Google Ads account; without active advertising spend, it only shows rough ranges rather than exact figures. Google Search Console provides the most valuable data of all free of charge: namely, the terms through which your own site is already being found, including impressions, clicks and average position. In the DACH region, Sistrix is a standard tool for visibility, competitive analysis and keyword data. In addition, AlsoAsked and AnswerThePublic provide the questions users ask about a topic, whilst Google Trends shows seasonal patterns and emerging trends.

3. Determine search intent

Every keyword is assigned a search intent. There are four types: informational (a thirst for knowledge), navigational (a targeted search for a brand or website), commercial (comparison before purchase, recognisable by modifiers such as ‘best’, ‘compare’, ‘test’, ‘alternative’ and ‘provider’) and transactional (specific purchase intent, recognisable by ‘buy’, ‘price’, ‘cost’ and ‘nearby’). The intent determines the format of the content: a guide for informational searches, a comparison table for commercial searches, and a product page for transactional searches. Find out more in the entry Search Intent.

4. Evaluate and prioritise

Now you evaluate each term along four axes: search volume, competition, business relevance and effort. Keyword Difficulty is a metric ranging from 0 to 100, which tools such as Sistrix, Ahrefs and Semrush estimate primarily based on the backlink profile of pages that are already ranking. A high value indicates strong, well-linked competition. Small and medium-sized enterprises in particular fare better if they do not target the highly competitive head keywords, but instead focus on more specific long-tail keywords with a clear intent.

5. Cluster and organise into a structure

In the final step, you group related terms into topic clusters. A cluster typically forms the basis for a pillar page with several subpages, each covering a specific aspect and its own search intent. This transforms a list of keywords into a well-thought-out information architecture that guides both users and search engines alike.

Practical example: Research using real tools

A concrete example from the B2B sector. A company offers migrations from Magento to Shopware. The obvious seed term is “Shop Migration”. Google Search Console shows that the page is already generating impressions for “magento shopware umzug”, but is ranked on page two. Sistrix adds related terms such as “magento 2 end of support” and “shopware migration costs”, along with search volume and difficulty. AlsoAsked provides the questions “How long does a shop migration take?” and “How much does a Shopware migration cost?”. This data does not result in a random list of keywords, but rather a prioritised content plan: a pillar article on migration, accompanied by specific subpages on costs, duration and data transfer, each tailored to the relevant search intent.

This approach combines first-party data (Search Console), market data (Sistrix) and questions from real users (AlsoAsked) to form a strategy based on demand rather than gut feeling. This is precisely the essence of professional search engine optimisation: content is created based on documented demand, not on internal assumptions. Google itself also recommends in the official SEO starter guide that you should create content for people and meet their actual needs.

Common mistakes in keyword research

• Focusing solely on volume. High search volumes are tempting, but without purchase intent they do not lead to conversions. A term with 200 searches and clear transactional intent often outperforms one with 5,000 searches that has no commercial relevance.
• Ignoring search intent. If you create a simple product page for a commercial term, you’ll fail to meet the searcher’s expectations and rank poorly.
• Underestimating the competition. If you don’t consider keyword difficulty, you’ll end up in a race against sites you won’t be able to catch up with in the foreseeable future.
• Treating research as a one-off task. Search behaviour changes, new terms emerge, and trends shift. Research is an ongoing process, not a project with an end date.
• Ignoring your own data. Many focus on external tools and forget about Google Search Console, which provides by far the most accurate and relevant data about your own site.

Frequently asked questions

How often should I carry out keyword research?

A comprehensive analysis is worthwhile at the start of a project and then at least once a quarter. You should also check Google Search Console monthly to discover new terms through which your site is already being found, and to respond to changes in search behaviour.

Do I need paid tools to carry out good research?

No, you can get started for free. Google Search Console provides your own performance data, the Google Keyword Planner gives rough search volumes, and Google Trends shows seasonal trends. However, paid tools such as Sistrix or Ahrefs provide more precise difficulty scores, competitive analyses and exact search volumes, which quickly justifies the cost for larger projects.

What is the difference between keyword research and SEO?

Keyword research is one step in the process of search engine optimisation. SEO also encompasses technical optimisation, content creation, link building and ongoing performance measurement. Keyword research provides the strategic foundation by determining which terms and topics all further SEO measures should focus on.

How many keywords should a page target?

A page should cover one clear main keyword and a few closely related secondary terms. Optimising a page for too many different terms dilutes its thematic clarity. It is better to create separate, focused pages for different search intentions.

Does keyword research still play a role when AI provides the answers?

Yes, even more so. AI systems and AI overviews derive their answers from indexed content. Those who rank well for the relevant terms and offer thematic depth will also be cited as sources in AI responses. The focus of research is shifting more towards questions, long-tail terms and topics with a clear intent, rather than pure volume.