Staff and customers were both avoiding the kiosk. The fix wasn’t the screen.
A mandatory ID kiosk was meant to speed up store visits. Instead, staff skipped it at peak hours and customers walked straight past it. I showed the team it was a trust problem, not a usability one, and the redesign we piloted halved peak queue wait.
SituationA kiosk everyone was quietly working around
A national telecom provider had rolled out a kiosk at the entrance of its stores. Customers had to verify their identity on it before speaking to anyone. New privacy regulation was pushing for stricter ID checks, and the internal hypothesis was simple: if customers identify themselves first, agents have less work and service gets faster.
That isn’t what was happening on the shop floor:
- At peak hours, agents bypassed the kiosk to stop queues from building up.
- Customers didn’t understand why they had to identify themselves before talking to someone.
- Older visitors needed help using it, which took an agent away anyway.
- Privacy-conscious customers refused to type personal details into a shared screen.
- Queue times weren’t improving, and every store enforced the policy differently.
TaskFind out why, and fix the right thing
I was brought in to improve the kiosk experience. Before planning any study, I rewrote the brief. Low adoption on its own told us nothing about the cause, and a usability study would only ever find usability problems.
How do we improve kiosk usability?
How might we design identification that balances operational efficiency, privacy trust and real retail behaviour?
That shifted the work from tweaking an interface to redesigning a workflow, and it widened who I needed to hear from: customers, store agents, store managers, retail operations and compliance.
ActionResearch the system, not the screen
01Mapped the whole identification journey
To see how policy, staffing and customer behaviour interacted, I mapped the service end to end, from walking in to leaving, across what customers do, what agents do, and the policy and systems behind them.
02Talked to people and watched the floor
Interview sessions used think-aloud on short kiosk tasks, so I could hear hesitation as it happened rather than a tidied-up version afterwards.
See the full research plan (11 steps)
03Found three teams measuring three different things
The stakeholder interviews explained the store behaviour. Each team was succeeding on its own metric, and nobody owned the customer’s experience in between. High kiosk adoption was being celebrated even when customers felt forced into it.
Pressure for speed meant agents skipped the kiosk whenever the queue grew.
One rule for every store, regardless of traffic or what the customer came in for.
A number that went up even when people used the kiosk reluctantly.
04Heard why customers resisted, and when they didn’t
People didn’t see why they had to check in before talking to someone, especially in a quiet store.
“If the store is empty, why should I enter my details? It’s faster to just talk to someone.”
Customers weren’t sure why their data was needed, how it would be used, or whether a public device was safe.
“I thought entering my phone number might mean getting marketing calls later.”
Showing ID to a person felt safer. The kiosk removed the human at exactly the moment people wanted reassurance.
“If I have to show ID, I’d rather show it to a person.”
When queues were long, people valued the speed and not losing their place, and many wanted to be notified instead of waiting.
“If it’s busy, this makes sense. I know I won’t lose my place in the queue.”
See the affinity map these came from
Customers also judged the kiosk against systems they already knew and trusted: token screens, a number on their phone, paper slips at government offices. None of those asked for personal details up front.
Identification isn’t a task. It’s a trust transaction. When a visit feels urgent or complex, people want a human. When it feels routine and predictable, digital works. The system failed because it assumed one context for everyone.
05Turned findings into a new way of working
Each problem became a specific recommendation, traced back to what people actually said.
- Dual-path workflow: self check-in for routine visits, an agent for complex or high-trust ones.
- Clear privacy messaging: say why ID is needed and how data is used, before asking for it.
- Peak-hour switch: managers turn kiosk-first on when the store gets busy.
- Assisted onboarding: agents introduce the kiosk to first-time and older visitors.
The biggest change was replacing one mandatory route with two. At peak hours the fast kiosk path is the default; when the store is quiet, customers choose between the kiosk and an agent, after a short explanation of why ID is needed.
Before: demand first, explain never
After: choice, reassurance, then a token
Store teams got the other half of the system: a live view of who was waiting and how busy the store was, so managers could switch the kiosk into peak mode when traffic rose.
ResultFaster stores, and a different definition of success
- Fewer manual overrides by agents, and smoother handling at peak times.
- Clearer communication about why ID is needed and how data is used.
- More voluntary kiosk use whenever the context suited it.
- The organisation stopped trying to force kiosk adoption and started fitting identification to the situation. I proposed measuring identification friction and trust clarity alongside adoption, so success meant customers feeling safe, not just using the screen.
ReflectionWhat I’d carry forward
Low adoption is a signal, not a verdict. Agents and customers weren’t being difficult. They were making sensible choices inside a system that didn’t fit their day. In operational products, the experience can’t be separated from the policy and workflow around it.
What I’d do differently: include lower-literacy users and non-English speakers from week one. Accessibility gaps surfaced late in this project, and I now plan for them from the start.
Full case study with all research artefacts and screens: meghanagautam.com/customer-identification.html · Visuals are AI-generated, depersonalised illustrations; the research and outcomes are real.
Client data is anonymised. To respect the telecom provider’s branding and intellectual property, all kiosk screens and store photos are re-imagined, depersonalised illustrations created with AI. The research, process and outcomes are real.