Chatbot Analytics: The 7 Metrics That Actually Matter
TL;DR: The seven chatbot metrics that matter are conversation volume, lead capture rate (13.6% across live assistants in Chatonbo's State of Ecommerce Chat 2026), resolution rate, average response time (under 5 seconds), conversation depth (4 to 8 messages), escalation rate (1.6% of conversations ask for a human in the same study), and after-hours engagement (44.7% of conversations in the same study). Check them in a 15-minute weekly review and a 30-minute monthly deep-dive.
Your chatbot dashboard is full of data - conversation counts, message volumes, response times, engagement rates. But which metrics actually matter for your business? Here are the seven you should track weekly.
1. Conversation Volume
What it is: Total number of conversations initiated per period.
Why it matters: Tracks adoption and reach. If volume is flat despite growing website traffic, your chatbot trigger settings may need adjustment.
Target: Growing month-over-month, proportional to your website traffic.
2. Lead Capture Rate
What it is: Percentage of conversations that result in a captured lead (name + email/phone).
Why it matters: This is your chatbot's primary conversion metric. It directly measures how well the bot qualifies and captures prospects.
Benchmark: 13.6% of conversations captured a lead across live assistants in Chatonbo's [State of Ecommerce Chat 2026](/research/state-of-ecommerce-chat-2026): 25.6% on service sites, 8.9% on online stores.
3. Resolution Rate
What it is: Percentage of conversations resolved without human escalation.
Why it matters: Measures the AI's ability to handle queries independently. A low resolution rate means your knowledge base needs improvement.
Target: Rising month over month as you fill knowledge gaps. Sales bots should escalate to humans more often for deal closing. For reference, [OMC](/customers/omc) reports 30% of its support tickets resolved by the assistant.
4. Average Response Time
What it is: Time between a user message and the bot's response.
Why it matters: Speed directly impacts satisfaction and engagement. Users expect near-instant responses from chatbots.
Target: Under 5 seconds for 95% of responses.
5. Conversation Depth
What it is: Average number of messages per conversation.
Why it matters: Longer conversations indicate engaged visitors. Very short conversations (1-2 messages) suggest the bot isn't capturing interest. Very long conversations (20+) may indicate the bot is failing to resolve or capture.
Target: 4-8 messages per conversation.
6. Escalation Rate
What it is: Percentage of conversations that require human handoff.
Why it matters: Too high means the AI isn't handling enough independently. Too low might mean it's not escalating when it should.
Benchmark: 1.6% of conversations ask for a human in Chatonbo's [State of Ecommerce Chat 2026](/research/state-of-ecommerce-chat-2026); watch the trend rather than a fixed target.
7. After-Hours Engagement
What it is: Percentage of conversations happening outside business hours.
Why it matters: This represents value you literally cannot capture without automation. These are leads and support requests that would otherwise be lost.
Benchmark: 44.7% of conversations started outside business hours in Chatonbo's [State of Ecommerce Chat 2026](/research/state-of-ecommerce-chat-2026).
How to Use These Metrics
Weekly review (15 minutes):
- Check conversation volume trend
- Review lead capture rate
- Scan for any conversations that should have been resolved but weren't
- Look for common unanswered questions
Monthly deep-dive (30 minutes):
- Compare all metrics month-over-month
- Review escalated conversations for knowledge base gaps
- Update knowledge base with new content
- Adjust trigger settings based on engagement data
FAQ
Which chatbot metrics should I track?
Seven: conversation volume, lead capture rate, resolution rate, average response time, conversation depth, escalation rate, and after-hours engagement. Review them weekly and do a deeper month-over-month comparison once a month.
What is a good lead capture rate for a chatbot?
In Chatonbo's State of Ecommerce Chat 2026, 13.6% of conversations on live assistants ended with a name plus an email or phone number: 25.6% on service and lead-generation sites, 8.9% on online stores. It is the chatbot's primary conversion metric.
What resolution rate and escalation rate should a chatbot have?
Resolution depends on how complete your knowledge base is; OMC, a Chatonbo customer, reports 30% of its support tickets resolved by the assistant. Requests for a human are rarer than most owners expect: 1.6% of conversations in Chatonbo's State of Ecommerce Chat 2026. Sales bots should still hand off to humans readily to close deals.
Written by
Sarah ChenHead of Growth · Chatonbo
Growth lead at Chatonbo. Writes about lead capture, conversion, and funnel design.
See it work on your own website
Paste your URL and chat with an AI agent trained on your content - ready in about 2 minutes.
Try It on Your WebsiteReady to try it yourself?
Deploy an AI chatbot on your website in about 2 minutes.
Start your 30-day free trial
