Once an AI receptionist or chat agent goes live, it's tempting to celebrate the number of conversations. Volume isn't the point. These seven metrics show whether it's working and what to fix next.
1. Answer rate
What it is: the share of inbound calls and chats that get a real response, rather than voicemail, hold or abandonment.
Why it matters: this is the first thing an AI agent should change. Compare it with the period before launch, especially for lunch hours and after hours.
2. Booking rate
What it is: of the conversations where someone wanted an appointment, how many ended with one booked.
Why it matters: it shows whether the agent is converting intent into business. A low rate usually points to limited availability, missing services or too many questions.
3. Hand-off rate and reasons
What it is: how often the agent passes to a person, and why.
Why it matters: some hand-offs are exactly right (complaints, complex questions). Repeated hand-offs for the same simple question mean the knowledge base needs an update.
4. After-hours share
What it is: the share of bookings and enquiries that come in outside business hours.
Why it matters: this is work you were most likely missing before. It's often the clearest evidence of value.
5. Speed to lead
What it is: time from a new enquiry (form, chat, missed call) to a meaningful response.
Why it matters: the faster you respond, the more likely you are to win the customer. With automation this should drop from hours to minutes.
6. No-show and reschedule rate
What it is: booked appointments that don't happen or get moved.
Why it matters: confirmations and reminders should bring this down. If it rises after launch, check that bookings are accurate and confirmations are going out.
7. AI search visibility
What it is: how often AI assistants (ChatGPT, Perplexity, Gemini, Google's AI Overviews) mention your business for the questions customers ask, and whether what they say is right.
Why it matters: this is the GEO side of the picture. Track a fixed list of questions monthly. Our guide to GEO explains how.
How to track them without a data team
- Your phone system, chat tool and booking system already record most of this; the work is putting it in one place.
- Start with a simple weekly summary: seven numbers, the change from last week, and one action.
- Read a sample of conversations every week. Numbers tell you where to look; transcripts tell you why.
AI analytics and dashboards are one of the six things we build, usually alongside an AI receptionist so the numbers are there from day one. Book a call if you'd like help setting them up.
Frequently asked questions
How soon should I expect to see results?
Answer rate and after-hours bookings show up in the first weeks. Booking rate and no-show improvements usually settle after a month of tuning.
What's a good booking rate?
It depends on your business, availability and the kinds of calls you get. Measure your own baseline in the first weeks and improve from there rather than chasing an industry average.
Do I need special software to track AI search visibility?
No. A spreadsheet with a fixed list of questions, checked monthly across the main assistants, works well to start.