Blog/How to measure an after-sales chatbot's ROI before buying
Guide

How to measure an after-sales chatbot's ROI before buying

The 4 metrics that matter (cost avoided, hours freed, NPS, resolution time) + a tested formula from the Mobilier de France case: from 2.5 days to 4 hours.

H'appi Team
February 28, 2026
8 min read

Why 90% of chatbot projects fail to prove their ROI

Most after-sales chatbot purchases are based on a vague promise: "you'll reduce costs." No baseline, no KPIs defined before deployment, no measurement methodology. Result: 6 months later, nobody can say whether it actually worked.

According to Gartner, 85% of enterprise AI projects fail to demonstrate measurable ROI within the first 18 months. This isn't a technology problem β€” it's a methodology problem.

The 4 metrics that actually matter

  • 1. Deflection rate: percentage of inbound calls/tickets handled by the chatbot without human intervention. Realistic target for a well-configured domain chatbot: 55–70%.
  • 2. Cost per resolved contact: total chatbot cost (subscription + maintenance) divided by the number of contacts resolved. Compare against the human cost (agent + infrastructure).
  • 3. CSAT (Customer Satisfaction Score): don't just measure whether customers got a response β€” measure whether they got the right one. A poorly configured bot can deflect 80% of contacts but satisfy only 30%.
  • 4. Average Handling Time (AHT): for cases that still reach an agent, the bot should reduce AHT by providing client context upfront. Target: βˆ’25% agent AHT.

The H'appi ROI calculation formula

Here is the formula we use systematically before any engagement:

  • Gross monthly savings = (monthly contacts Γ— deflection rate Γ— avg duration) / 60 Γ— agent hourly cost
  • Net monthly ROI = gross savings βˆ’ chatbot subscription cost
  • Payback period = deployment cost / net monthly ROI

This formula is deliberately conservative: it doesn't account for dispute reduction, customer satisfaction improvements, or savings from extended hours (bot available 24/7).

Real-world case: 300 calls/month at 28 €/h

  • 300 after-sales calls/month Γ— 65% deflection = 195 automated calls
  • 195 Γ— 8 min / 60 Γ— 28 €/h = 728 € saved per month in pure labour
  • H'appi Bot subscription: 299 €/month
  • Net ROI: +429 €/month from month one, or +5,148 € over 12 months
  • Not counting dispute reduction and 24/7 availability

Our online ROI calculator lets you adapt these figures to your real situation in under 2 minutes.

Mistakes to avoid in your calculation

  • Overestimating deflection rate: a vendor promising 90% deflection without seeing your data is misleading you. Use 55–65% for an honest calculation.
  • Forgetting implementation time: 2 to 4 weeks during which your team is mobilised. Include this in year-1 ROI.
  • Not measuring post-bot CSAT: a bot that deflects but frustrates can cost more in customer churn than it saves in calls.
  • Comparing against zero cost: the status quo has a cost. Measure it honestly before comparing.
How to measure an after-sales chatbot's ROI before buying β€” H'appi Blog | H'appi