Why workshop WhatsApp becomes difficult to manage

Customers often message while service advisors are speaking with another customer, checking a vehicle or coordinating technicians. The same inbox may contain tyre enquiries, service requests, appointment questions, quotation follow-ups, complaints and technical questions.

The opportunity is not to let AI answer everything. It is to separate routine, approved questions from requests that need a workshop employee and to make the customer information usable after the conversation.

Customer messageIdentify requestApproved responseCapture detailsHuman handover if needed

A controlled WhatsApp workflow

WhatsAppIntentKnown detailsKnowledge baseCRM / staff
  • Acknowledge the customer and identify the reason for the enquiry
  • Collect approved details such as name, mobile number, vehicle registration, make, model and requested service
  • Answer only from the workshop-approved knowledge base
  • Ask only questions that are still missing instead of repeating information the customer already provided
  • Escalate diagnosis, safety, complaints, discounts, uncertain pricing and custom work to an authorised person
  • Create or update the CRM record and next action
  • Trigger approved booking, quotation or follow-up workflows after the conversation state is known

What changes at 100, 500 or 1,000 enquiries per month

These are planning examples, not performance promises. At 100 enquiries per month, a small workshop may mainly need consistent capture and reminders. At 500 enquiries, assignment, status and exception visibility become more important. At 1,000 enquiries, routing by outlet, service type or staff owner may become essential.

The correct automation design depends on the enquiry mix, how many conversations need human judgement and whether the workshop already uses a CRM, booking tool or workshop-management system.

  • 100 enquiries/month: focus on consistent intake and missed follow-up prevention
  • 500 enquiries/month: add clear assignment, status and escalation reporting
  • 1,000 enquiries/month: consider outlet/service routing, queue controls and structured reporting

When the AI should stop and hand over

Where the connected platform permits, the customer can stay in the same conversation. Once a staff member takes ownership, the AI should remain silent until control is deliberately returned.

  • The customer asks for a person
  • A safety concern or possible vehicle failure is mentioned
  • Technical diagnosis or repair recommendation is required
  • Pricing, warranty, insurance or discount information is not explicitly approved
  • A complaint or reputationally sensitive issue is received
  • The request is custom, unusual or outside the approved knowledge base
  • The system has insufficient confidence to answer safely

What to measure in a pilot

  • Enquiries received and correctly classified
  • Percentage answered from approved content
  • Escalation rate and escalation reason
  • Vehicle and customer records captured completely
  • Booking or quotation requests created
  • Follow-ups completed on time
  • Staff corrections and exceptions

Important note about the examples

Workflow examples, enquiry volumes and document-volume calculations in this article are illustrative and are provided to explain possible automation designs. Actual integrations, processing time, extraction accuracy and operational outcomes depend on source quality, business rules, connected systems, APIs, account access and human-review requirements. Performance figures are only presented as tested results where specifically identified.