Semantic search on otto.de: synonyms instead of keyword stuffing

What OTTO says about semantic search

At the end of November 2024, OTTO Advertising introduced an AI-based semantic search. According to retail-news.de, it doesn't just match individual search terms but recognises their meaning and possible synonyms. OTTO's example: someone searching for "Jogging-Schuhe" (jogging shoes) also gets results for related terms such as "Laufschuhe" (running shoes) or "Sportschuhe" (sports shoes).

OTTO says this improves both organic search results and the delivery of Sponsored Product Ads. For ads it gives figures: more ad slots per results list (from four to eleven) and a 26 % higher click-through rate. For organic search, OTTO publishes no figures.

Why keyword stuffing makes even less sense now

The temptation used to be to cram every variant into the copy: "running shoes jogging shoes sports shoes trainers". If search maps synonyms itself, a list like that adds no new information. It does make the listing harder to read – for shoppers in the results list and on the product page.

Semantic search doesn't replace clean product data, though. It can connect "jogging" with "running", but it can't know your shoe is waterproof if that isn't stated anywhere.

What still matters in the title

On OTTO you don't write the title freely: according to the OTTO Market API's category interface, each category group shows "how our product title is built". You influence it through the fields it is built from – category, brand, product line and attributes. That means:

• The right category sets the product type in the title. No amount of synonym matching rescues a wrong category.
• The product line (at most 70 characters according to the API, identical for all variations) is a proper name, not a keyword list.
• Attributes such as size, material or colour supply the features shoppers search and filter by.

More in Optimising product titles for otto.de.

Where synonyms fit naturally

Synonyms don't belong in a list, but they can appear where they come up naturally. According to the API documentation, the description is "relevant for SEO" and should stay under 2,000 characters. A sentence like "This lightweight running shoe suits jogging laps in the park and short races" uses two terms and says something about use at the same time.

Bullet points (3 to 180 characters each) stay reserved for concrete features. And the wording in reviews and customer questions shows which words buyers actually use – more in Finding long-tail keywords on OTTO.

Testing: track the main term plus 2–3 synonyms

OTTO doesn't publish whether semantic search finds your product as well for synonyms as for the main term. You can check it for your own products:

1. Pick the main term per product (e.g. "Laufschuhe Herren") and 2–3 synonyms shoppers use ("Joggingschuhe Herren", "Sportschuhe Herren").
2. Record the organic position for all of them over one to two weeks so ads don't distort the picture – see Overall, organic and ad position.
3. Compare: if the main term and the synonyms sit close together, search already connects the terms for your product.
4. If one synonym ranks clearly worse, add the term at a natural point in the description, or check whether a relevant attribute is missing.
5. Change only that, note the date and compare over several weeks.

Reading the results

A synonym can rank worse for good reasons: shoppers using that word may be looking for something else, and search rightly shows other products. Look at the first results page for the synonym before changing anything. If it shows entirely different product types, the term isn't the right one for your product.

Also expect normal fluctuation. Only when the position moves out of its usual range over several weeks is a change confirmed. The method in detail: Measuring rankings after listing changes.

Semantic search and AI advice

The same shift shows up in OTTO's in-app AI assistant and the ChatGPT plugin: shoppers describe needs in full sentences rather than keywords. According to t3n, the assistant also understands dialects and colloquial language. Complete attributes and clear descriptions help in both worlds – see OTTO in ChatGPT and the AI assistant. A rank tracker such as MarktRank shows you the organic position separately for each term.

Frequently asked questions