71% of tickets, no human touch
A direct-to-consumer home-goods retailer watched support volume track revenue. First response averaged six hours in peak weeks. Where-is-my-order and returns made up two thirds of all volume. A rules-based chatbot had already failed, because it could talk about orders but could not look at one. The new build connected live order, tracking and customer data before any prompt work. Return and refund rules sit in the system as hard limits: value caps, day windows and category exclusions. Inside the limits the agent acts. Outside them a person approves with one click. Launch ran two weeks in shadow mode, with the agent drafting and people approving, then opened category by category.
Start with one job only. Fill in the three lines. Keep the bot on that job.
The one job this bot does: ______________________________
What it may do on its own: ______________________________
What it must ask me first: ______________________________
Two lines from this case, side by side. No new figures.
A direct-to-consumer home-goods retailer watched support volume track revenue. First response averaged six hours in peak weeks. Where-is-my-order and returns made up two thirds of all volume. A rules-based chatbot had already failed, because it could talk about orders but could not look at one. The new build connected live order, tracking and customer data before any prompt work. Return and refund rules sit in the system as hard limits: value caps, day windows and category exclusions. Inside the limits the agent acts. Outside them a person approves with one click. Launch ran two weeks in shadow mode, with the agent drafting and people approving, then opened category by category.
Within 8 weeks of full launch, 71% of inbound tickets closed with no human touch. First response fell from a 6-hour average to 9 seconds, and held through a Black Friday volume spike of 3.4 times. Satisfaction on automated resolutions ran 18 points above the old human baseline. Planned peak-season hiring was cancelled. The retailer is not named, and the figures are the agency's.
The after line is the source's claim, not our measurement.
smalloffice.space · Get a case like this in your inbox each morning. Subscribe free
Read this case on your phone now, and keep it for later. Type your email. We send this case to that address, and you get the next cases each morning too.