The Ticket Isn't Closed Until Confidence Is Repaired

The Volley trainer is an AI machine that launches balls at up to 80 mph and serves from as high as 84 inches. It is safe, and customers know it is safe. But when there is an error, even the slightest one, they prefer to stand on the opposite side of the court from it.

That reluctance is expensive. A connected device can often come back quickly if the customer can describe a few basic things they see. A customer who keeps their distance says little, so the unit stays down longer and is not available for the next customer. For Volley, that can mean a trainer out for a day instead of an hour. Sometimes it means swapping the machine, which means crating a 250-pound unit and shipping it back and forth from the distribution center.

Reluctance outlasts the repair. Confidence in the machine does not return just because the machine is fixed. A ticket can close with resolution time under target and a neutral survey, and still be a false positive on confidence. The reluctant customer quietly uses the product less. The dashboard shows a machine that works, while the reluctance seeps into the renewal decision.

A New Kind of Failure

A vending machine fails in ways everyone understands. A machine that moves on its own, makes decisions and does things customers find impressive fails in ways they cannot predict. Customers arrive a little awed and a little on their back heel, unsure how to engage.

Most of them have never used anything like it. They have no baseline for what a normal failure looks like, so every failure looks like the machine misbehaving. Awe turns into wariness.

Researchers call this perceived safety: How dangerous an interaction feels, and how comfortable a person is during it. In a NIOSH experiment, people reported more fear as a virtual robot arm moved faster. In a separate construction study, separating human and robot work areas raised perceived safety, and the safer people felt, the more willing they were to work with the robot again.

Standing across the court is the same instinct. The studies involved workers and robot arms, not customers and consumer machines, so read them as a clue about the pattern, not proof of the size of the effect.

Repair One: The Machine

The technical fix comes first, and it has to hold. A University of Michigan study had 240 participants work with a robot colleague that made errors and tried to repair trust. After three mistakes, none of the repair strategies fully restored it. The researchers concluded that a robot needs to be sure it has mastered a task before it tries to repair a human's trust.

A 2026 study points the other way, with a catch. A success right after a single failure had the same effect on perceived reliability as a success with no failure before it. Read together, one failure followed by a clean result is recoverable. A repeat failure is the dangerous one. Verify the fix before the confidence conversation starts.

Repair Two: The Confidence

Confidence repair has three parts, and none of them is optional.

Explain the failure plainly. An opaque failure is scarier than one the customer understands. "The sensor misread the ball in low light, we recalibrated it, and here is what you will see if it happens again" gives a customer something to hold onto. A technician who says it is fixed without saying what it was leaves the customer to invent an explanation. The invented one is usually worse.

Show it working. Run the product with the customer watching before anyone leaves. In the field, it looks like this: The technician recalibrates the sensor, then asks the customer to launch the first ball. A customer who watches a fix has seen a demonstration. A customer who runs it and sees it work has their own proof. It worked for them, so it can work for them again.

Show what is better. The goal is not to get back to the old baseline. The customer should leave believing the machine is at least as good as it was, ideally better. Say what changed, what was updated and what will now catch this earlier.

When No One Is There

Much of this work happens without a technician in the room. Fixes go out over the air. A local handyman talks a customer through a repair on FaceTime. At Volley, 200 robots ran nationwide with no field service staff.

Because no one was in the room, we had to be present another way. Customer management, customer support and customer success all worked to make sure the customer knew we were with them the whole time, even though we were not there as staff. Four things made that real. Our machine diagnostics gave us a lot of information on the likely error. Our support app was built around communication. We reached out to customers first instead of waiting for a reaction. And we updated our clubs on usage issues on a steady, frequent cadence, so they always felt in the loop.

The three parts of confidence repair have to travel to wherever the customer is. A plain explanation on the screen or in the app. A visible successful run after the fix. A short note on what changed.

Measure It

Time to resolution tells you the technician left. It does not tell you the customer came back. Return-to-use time, usage in the weeks after an incident and a one-question confidence check at ticket close belong on the same dashboard as resolution speed. If usage has not recovered three weeks after a fix, the ticket closed early, whatever the other figures on the dashboard say.

The obvious objection is that customers want it fixed fast, and extra explanation signals a fragile product. Speed comes from the technical repair. The confidence step is a few minutes, and skipping it bets that the customer will work it out alone.

The Definition of Done

Closing the ticket repairs the product. Repairing confidence keeps the customer. At VS Networks, where the fleet grew from 1,000 to more than 20,000 kiosks, churn stayed under 1 percent. Trust was a main driver in that result. We shaped our customers’ perspective that, while machines naturally have issues, they could count on us to bring them through the issue and back to day-to-day as soon as possible.