OTTO in ChatGPT and the AI assistant: getting your product data ready

What OTTO has announced

According to retail-news.de (20 October 2026), OTTO has launched a plugin for ChatGPT. Shoppers install it from ChatGPT's plugin section and start a consultation with "@OTTO". They describe what they need and get matching products from OTTO's range – the reports mention household appliances and coffee machines as examples. OTTO says the advice covers more than 19 million items.

The purchase itself still happens on otto.de. Under the hood is an interface based on the Model Context Protocol (MCP), built by the group's own technology partner one.O.

OTTO also says it has been testing ads in ChatGPT since September 2026 as one of the first European partners. It has not published any results.

What doesn't work yet – and what's unclear

An important caveat: according to etailment, OTTO explicitly describes AI agents checking current prices and stock or placing orders as future possibilities, not as features available today. There is no timeline.

It is also unclear whether marketplace partners' offers are part of the plugin's catalogue. The reports refer to "OTTO's range" without saying whether that includes partner offers (as of October 2026). So don't count on traffic from ChatGPT – but get your data into a state that works for any kind of AI-led advice.

The AI assistant in the OTTO app

Since March 2026 the OTTO app has had a shopping assistant that OTTO developed with Google, based on Gemini. Shoppers ask by text or voice, the assistant asks follow-up questions and suggests products. t3n gives examples such as an Italian espresso machine with a portafilter, or a vacuum cleaner for a flat where someone has a house-dust allergy.

According to t3n, the assistant uses OTTO's own data, including product descriptions and structured attributes; reviews are to follow later. OTTO reports up to 37 % higher basket values and twice the conversion rate of classic search – figures that, according to the reports, apply only to its own in-app assistant, not to the ChatGPT plugin.

How AI advice picks products

In classic search, a shopper types "coffee machine". Talking to an assistant, they are more likely to write: "quiet coffee machine for a small kitchen, under 200 euros". The AI has to turn that into features – noise, dimensions, price – and find products whose data matches.

That only works if those features are in your data. If the width isn't in the attribute and suitability for small kitchens appears nowhere in the text, the AI has no reason to suggest your product. OTTO hasn't published how it weights the data. What is certain: anything missing can't be matched.

1. Complete and correct attributes

Attributes are the structured foundation – for filters and, according to t3n, for the assistant too. Fill in not just the mandatory attributes but every one that fits the category: dimensions, material, power, colour, target group. Use the recommended values from the category interface, the right unit and, for multi-value attributes, every value that applies. Details in Categories and attributes on OTTO.

2. Describe use cases in shoppers' language

The description is the place for what no attribute captures: who the product is for and in which situation it helps. Write concrete sentences such as "at 18 cm wide it fits narrow worktops" – provided it's true. Generic lines like "ideal for every household" help neither shoppers nor an AI.

• For whom? (single household, family, allergy sufferers, beginners)
• For what? (office, camping, small flat)
• What sets it apart? (quiet, compact, washable – with a number where possible)

More on structure: Writing product descriptions for OTTO. If you draft copy with AI, read Writing OTTO product copy with ChatGPT first.

3. Keep an eye on stock and reviews

Stock and price checks by AI agents are still in the future at OTTO. It is still worth reporting stock precisely now: a recommended product that can't be delivered disappoints the shopper, and out-of-stock products can also drop in regular search. See Stock and availability on OTTO.

According to t3n, OTTO plans to feed reviews into the assistant. Answer customer questions and read reviews for clues about which features matter to buyers – those are often exactly the phrases they will use with an AI.

What you can measure

How often ChatGPT or the app assistant recommends your product can't be measured from the outside. What you can measure is classic search on otto.de, which according to OTTO has also worked semantically since the end of 2024 – see Semantic search on otto.de. Better data should show there first: track your products' organic position for the main term and for typical needs-based phrases, for example with MarktRank, and compare before and after you clean up the data.

Frequently asked questions