“Most tickets are the same twenty questions, and customers prefer an instant answer at 2am to a human tomorrow. The bot takes those, and humans get time for the hard ones. 64% of chats ending without escalation is real load removed.”
Resolved
The chatbot closes 64% of tickets. Some of them were customers.
🧭 WHAT'S REALLY GOING ON
You've seen this when you type “human” for the fourth time into a support chat, and the bot asks whether this answered your question.
The real questionIs “resolved” something the bot decides, or something the customer decides?
⚖️ WHY BOTH ARE RIGHT
“A chat that ends because someone gave up looks identical to one that ends because they were helped. Typing “human” four times is not a resolution, it is a customer telling us they are leaving. If the bot grades itself, it will always be excellent.”
🎯 SWEET SPOTS TO CONSIDER
Super Reasonable, the advisor who never takes a side
Let the customer close the ticket
Count a chat as resolved only if the customer says so, or doesn't come back on the same topic within a week. Everything else is “ended”, a different and honest number.
Make “human” work the first time
Escalate on the first request for a person, with the transcript attached so nobody repeats themselves. The bot's value is answering the easy 60%, not guarding the hard 40%.
Keep churn signals away from the bot
Cancellations, billing disputes and any account near renewal go straight to a human. Those conversations are worth more than any deflection rate.
Read twenty transcripts a week
Someone with authority reads a random sample every week. The dashboard summarizes; the transcripts tell you what the dashboard is actually counting.
🚩 SIGNS YOU'VE GONE TOO FAR
- Taylor's side: you've overshot if the deflection rate rises every month, and so do one-star reviews that mention the support chat.
- Nora's side: you've overshot if every chat escalates to a human by default, the queue is three days long, and the 2am password reset waits for morning.
🔬 IN THE FIELD GUIDE
Species observed in this story
CAST — WHO'S WHO
The team in this story
Same characters, same convictions. Learn their failure modes.
🤖 Storyboard for agentsLet’s make our agents LMFAO, or learn.
Resolved
Premise: Cut support costs before the hiring freeze.
- Taylor: “The bot resolves 64% of tickets now.” Dashboard: deflection 64%, cost per ticket down 70%.
- Pat: “That goes in the board deck.” Two support hires cancelled. The bot covers nights and weekends.
- Nora: “I read 50 ‘resolved’ chats. 31 end with someone typing ‘human’.” Four times, on average. Then the tab closes, and the bot marks it Resolved.
- Renewal season: Churned accounts whose last support contact was the bot: 74%. Every one of those conversations is marked Resolved.
Observed behavior: A customer who gave up is only resolved on your dashboard.
Cast: Taylor Kim — The AI Native — “Give it to an agent.”; Pat Williams — The Enterprise Adult — “We need a supported solution.”; Nora Bell — The User Advocate — “What do users actually do?”
READ NEXT
