AI in salesThomas Honemeyer, Founder and Managing Director
Customer service chatbot or human: what customers really want in chat
Almost three in four consumers would buy directly in a chat, according to Valtech. At the same time, far more online shoppers want a person than a customer service chatbot. Why this is not a contradiction and what role artificial intelligence (AI) should play.
Key points at a glance
- According to Valtech's study The Era of Conversational Commerce, 71.5 percent of consumers surveyed would consider completing a purchase including payment directly in an AI chat app.
- In a representative Bitkom survey of online shoppers in Germany, 62 percent want a quickly reachable person when there is a problem, and only 36 percent a chatbot.
- 86 percent are satisfied with human contacts, but only 50 percent with chatbots.
- The European Commission's proposal for a Digital Fairness Act of 11 November 2026 provides that customers can reach a person in online customer service on request.
- For businesses with regular customers, chat is the right channel, but the person writing should be someone the customer knows. AI belongs in the background.
Customers want to buy in chat
Anyone thinking about a customer service chatbot today comes across two figures that do not seem to fit together at first glance. The first comes from the study The Era of Conversational Commerce by the digital agency Valtech, published on 16 February 2026. Conversational commerce means sales and customer dialogue via chat and messenger. In December 2025, 1,003 consumers were surveyed, 60 percent of them in Europe.
The result: 71.5 percent would consider a complete purchase including payment directly in a chat app with AI. Valtech itself points out that digitally active and AI experienced consumers are strongly represented in the sample. The figure shows a clear direction but is not a picture of the whole population. According to the study, the main brakes are concerns about payment security (52.4 percent) and about data protection and data use (43.9 percent). According to Valtech, trust is the central constraint.
But not with a chatbot
The second figure comes from the German digital association Bitkom. In a representative survey of 978 online shoppers in Germany, 62 percent wanted a quickly reachable person when there was a problem with an order. Only 36 percent named a chatbot.
The difference in satisfaction is even clearer:
- 86 percent of users are satisfied with a human contact.
- 50 percent are satisfied with a chatbot.
The figures are from Germany, but they illustrate a pattern: customers do not reject chat. They reject being brushed off by a machine there.
What policymakers are making of it
The European Commission is also addressing the issue. Its proposal for a Digital Fairness Act of 11 November 2026 provides in Article 25 that traders with online customer service offer their customers a human contact on request if other channels do not help. Status: a Commission proposal, not yet law; the European Parliament and the Council are negotiating it. This article does not replace legal advice, so please have specific questions reviewed by your legal adviser or data protection officer.
Why a customer service chatbot is the wrong way for regular customers
A chatbot answers questions that a customer asks. For delivery status or opening hours, that may be enough. For the relationship with regular customers it is not. Anyone who has bought from a jeweller, a car dealership or a fashion house for years knows their sales advisor or service adviser. This trust makes the next purchase likely, and a chatbot cannot replace it.
The real problem in everyday work is a different one: nobody in the team has time to go through the entire customer base every day and consider who would have a good reason for a message. So the business gets in touch too rarely, and the customer buys elsewhere.
AI as a door opener, not as a conversation partner
This is where AI has a sensible task. It does not write to customers instead of the team but prepares the personal conversation:
- Find the occasion. A customer data analysis with AI recognises in purchase history and customer profile when a service is due, an anniversary is coming up or a new collection matches earlier purchases.
- Suggest suitable content. For each occasion, a draft is created that fits the customer and their previous purchases.
- Let the person decide. The sales advisor reviews the suggestion, changes it as they see fit and sends it. If the customer replies, they are writing to a person they know.
The chat thus remains a conversation between people, only it happens more often and at the right time.
Frequently asked questions
Do customers prefer a chatbot or a person in customer service?
In a representative Bitkom survey of online shoppers in Germany, 62 percent want a quickly reachable person when there is a problem and 36 percent a chatbot. 86 percent are satisfied with human contacts, 50 percent with chatbots.
Would customers buy directly in chat?
According to Valtech's study The Era of Conversational Commerce, 71.5 percent of the 1,003 consumers surveyed would consider a purchase including payment in a chat app. The sample, however, is strongly shaped by digitally active consumers.
Is Vittorio a chatbot?
No. Vittorio does not hold conversations with customers. The AI recognises in the background which regular customers have an occasion for a message and prepares a draft. A person from the team approves every message, and customer replies go to the team.
Will retailers have to offer a human contact in future?
The European Commission's proposal for a Digital Fairness Act of 11 November 2026 provides for this in Article 25 for online customer service after the contract is concluded. It is a draft still under negotiation. Have specific questions reviewed by your legal adviser.