WhatsApp Business and chatbots: what does your customer service need?

App, platform and chatbot: three different concepts

Official documentation describes quick replies, greeting messages and away messages in the WhatsApp Business app. The platform provides interfaces for integrating messaging with other systems. A chatbot is the conversational logic configured for particular use cases; its behaviour depends on the implementation. Product features and availability can change.

Reference: WhatsApp Business App · WhatsApp Business Platform

What conversational research contributes

Adam, Wessel and Benlian (2021) analysed an experiment with 153 German participants in the final sample. Certain human-like interaction cues increased willingness to accept a chatbot request. The outcome was agreement to provide service feedback, not sales or issue resolution, and the study did not evaluate WhatsApp. Its relevance here is that interaction design deserves its own evaluation.

Reference: Adam, Wessel & Benlian, 2021

Start with an enquiry inventory

Review a sample of conversations without transferring unnecessary personal data. Group contact reasons and identify when staff must consult an application, confirm information or handle an exception.

• Stable information: opening hours, location and services.
• Changing information: availability or enquiry status.
• Actions: recording, confirming or changing appointments.
• Special cases: complaints, ambiguity and out-of-scope requests.

A correct answer needs a source and a boundary

Identify who maintains each information source and how it is updated. Define what happens when data is missing. An assistant should acknowledge that it cannot confirm availability when it has not checked the booking system.

Hypothetical example: a customer asks to move an appointment to Friday. Without authorised calendar access, the assistant can collect a request for review but cannot confirm the change. With integration, confirmation must depend on the calendar response and agreed controls. This illustrates a design, not a documented ALKATECH deployment.

Design human handover before launch

Identify the assistant as such and provide a clear way to request human help. Specify who receives the case, what context is passed on and how the customer knows it is pending. If nobody is available, explain the actual next step.

Test misspellings, repeated requests, language changes and integration failures. Include cases where the correct behaviour is to stop an action or ask for clarification. These are practical design recommendations, not effectiveness estimates from the study.

Measure resolution as well as speed

Define a resolved enquiry using verifiable criteria: for example, a correctly recorded confirmation accepted by the customer. Track errors, reopened cases, abandonment and handovers alongside response time. Do not classify every conversation without human involvement as a success.

A first project can cover a limited set of common questions. Evaluate it before adding integrations or actions. Costs, requirements and data handling should be specified for the chosen solution.

Frequently asked questions