How to measure whether your website chatbot is paying for itself
A chatbot is worth keeping if it brings in or saves more than it costs. You don't need analytics software to check: a month of numbers from your dashboard and a rough idea of what a customer is worth is enough.
1. Count the leads it captured
Look at the leads your chatbot collected in a month (name and phone or email, plus what they asked for). Then check how many turned into paying customers. Even a rough count from your own records is fine.
2. Estimate the time it saved
Count the conversations it handled and how many were simple questions (hours, prices, location, how to book). Multiply by the minutes each would have taken your team on the phone or WhatsApp.
3. Put it together
- Value from new customers = customers who came through the chatbot × average first purchase
- Value of time saved = hours saved × what an hour of your team's time costs
- Return = (value from new customers + value of time saved) − monthly cost of the chatbot
Example: 3 new customers × ₹2,000 + 10 hours saved × ₹200 = ₹8,000 a month, against a ₹3,499 plan. If one extra customer a month covers the cost, the decision is easy.
4. Look at what it couldn't answer
Questions the chatbot couldn't answer are lost leads waiting to be recovered. Add the common ones as FAQs each week; the share of questions it answers well should go up month by month.
Where to find these numbers in ONNT
The dashboard shows AI replies and leads per day (Overview), every conversation for the last 90 days (Conversations), captured leads with a CSV export (Leads), and the questions it couldn't answer.