Keyword research

Keyword research finds out which words your customers use to search for what you sell, and how many of them do so. It is the foundation for category structure, content and Google Ads campaigns. Skip it, and you build a store in the merchant’s language instead of your customers’ language.

Reviewed by Zoran Artmagic · November 6, 2026

Merchants and customers rarely speak the same language. In the store, the category is called “technical jackets”; people search for “breathable rain jacket women”. The manufacturer calls it “ergonomic seating”; the customer types “office chair for back pain”. Keyword research closes exactly this gap. It systematically collects the terms people search with and shows how often, and with what intent, they do so.

Research and analysis belong together, but they are two steps. Research collects and evaluates terms. Keyword analysis then checks how well an existing page matches its terms and where it needs sharpening.

Why keyword research decides revenue

Every page, every category and every campaign targets search terms, whether deliberately or not. Choose the wrong ones and you optimise neatly in the wrong direction: a lot of work, few suitable visitors, hardly any orders. In Google Ads, you even pay for every single click on the wrong terms. Good research prevents this before the first euro and the first working hour are spent.

Keyword research in five steps

Step 1, collect ideas: The best sources are in-house. Your product range, the search in your own store, questions from customer service and sales, the words in product reviews and the search terms report from Google Ads. Add a look at how competitors name their categories.
Step 2, expand with tools: Google Keyword Planner provides related ideas with search volume for a starting term. Google Search Console shows which terms your store already appears for. Tools such as Sistrix, Semrush or Ahrefs also show what competitors rank for. The entry on keyword tools gives an overview.
Step 3, add questions and the long tail: Google autocomplete, the “People also ask” section on the results page and tools such as AnswerThePublic supply the longer, more precise searches. AI assistants are good for generating ideas. Their search volume figures, however, are guesses and do not belong in the evaluation.
Step 4, evaluate: Every term needs search volume, competition and, above all, search intent. Does someone want to buy, compare or learn? A look at the results page shows whether Google displays categories, products or guides for it. Check seasonal fluctuations with Google Trends.
Step 5, group and assign: The terms are grouped by topic. Whatever has the same intent belongs on the same page. The result is a keyword map: every group of terms has exactly one target page, either an existing one or one that still needs to be created.

Search volume is not the same as revenue

The most common mistake in evaluation is looking only at search volume. For a merchant, something else counts: how likely is someone who searches like this to buy, and what is the purchase worth?

An example: “running shoes” is searched very often, but with unclear intent and against large competitors. “waterproof trail shoes women wide feet” is searched only a few hundred times a month. But whoever searches like this knows exactly what they want and is close to buying. With a basket value of €150, the small term often brings more revenue than the big one, for a fraction of the effort.

On top of that, search volumes are estimates. Without running campaigns, Keyword Planner shows only rough ranges, and every tool calculates differently. What matters are the orders of magnitude, not the exact figure.

Where the terms belong in your store

Category pages: the main term with its variants, such as “trail shoes women”. In a store, this is usually where the greatest potential lies.
Filter pages: only as a separate, indexable page if there is real demand for the combination, such as “waterproof trail shoes women”. All other filter combinations stay invisible to Google, otherwise you create duplicate content.
Product pages: model names, item numbers and very specific purchase searches.
Guides and encyclopedia: the questions before a purchase, such as “do trail shoes run small”. They bring visitors at an early stage and link to the matching categories.

Every page gets its own topic. If two pages compete for the same terms, they weaken each other. How a group of terms becomes a page that sells is described in the entry on the landing page.

Keyword research in the age of AI search

In ChatGPT, Claude, Gemini or Perplexity, people ask whole questions instead of typing keywords. This does not make keyword research less valuable; it makes it broader. Alongside individual search terms, topics and the questions customers ask around a product count. Whoever knows these questions and answers them clearly on their pages is found in Google Search and named in AI answers. More on this under Generative Engine Optimization.

The most common mistakes

Following search volume only: The big terms are fiercely contested and often unclear in intent. Revenue is frequently made in the small ones.
Thinking in your own language: Hardly any customer searches for internal names and manufacturer terms.
Researching once and never again: Product ranges grow, trends come and go, customers change their words. The list is alive.
Terms without a target page: A keyword list that isn’t assigned to any page remains a spreadsheet and doesn’t bring a single visitor.
Overlooking your own data: Store search, Search Console and the search terms report show what real customers actually type in. No tool knows your customers better.

How the research turns into a strategy for your store is shown on SEO for online stores. The same applies to paid search under Google Ads for Shopware and Shopify.

Keyword research translates your product range into your customers’ language. Skip this translation, and you talk past them, no matter how well everything else is done.