Chief Product Officer / India
What visitors revealed about our product decisions
- Hospital organisations
- 36
- Customers of nonclinical visitor-information products
- Product, design and research colleagues
- 18
- Three leads; engineering reports separately
- Information-design experience
- 29 years
- Printed service information to hospital wayfinding products
Decisions & consequences
01Changing a map that was accurate but pointed people the wrong way
In a 2022 kiosk exercise, seventeen of forty participants chose the wrong initial direction after reading a floor plan. The plan accurately showed the building, but the display faced the opposite way from the visitor's orientation. I had approved north-up maps throughout the range because a uniform presentation looked consistent and simplified our design system. The exercise showed that the consistency helped us more than it helped the person deciding which corridor to take.
The visual-design team proposed adding a more prominent compass. Site colleagues argued that many visitors were trying to relate the screen to the space in front of them, not determine a geographic bearing. I chose a placement-aware orientation approach with a clear position cue and installation review. Hospital estate owners retained approval of the route information. We did not infer that rotating a map alone established step-free access, safety or suitability for every visitor.
The redesign cost ₹28 lakh over eight weeks and postponed the planned three-dimensional map feature. Commercial colleagues were disappointed because that feature was prominent in demonstrations. I explained why a visitor's first turn should take priority over a more impressive presentation. Nisha Dervqeli led the repeat exercise with forty participants under the defined conditions. Thirty-five selected the intended first direction and five still required assistance. We kept the unresolved cases for further study rather than describing the product as independently usable by everyone.
The result changed the installation requirements as well as the screen. A map's orientation had to be reviewed in relation to its placement, and later changes to the kiosk position could require another check. The reported outcome was the observed task result, not a reduction in missed appointments or an improvement in health outcomes.
I learned that product consistency should protect the meaning of an interaction rather than force identical geometry everywhere. The original plan was not geographically false, but it did not answer the visitor's immediate question well enough. That distinction helped me explain why an apparently small configuration detail belonged in the product's definition of a successful installation.
Replacing an internal abbreviation with an agreed destination model
A 2024 research exercise asked thirty participants to find the destination described in a supplied hospital instruction. Twelve could not connect the public department name with the internal abbreviation used by the kiosk. Both labels referred to the same destination, but the product expected visitors to know the organisation's vocabulary. I had treated destination naming as customer content and had not given the product a sufficient way to preserve the relationship across outputs.
The first proposed correction was to enlarge the abbreviation and add a search hint. That would make the existing label more visible without resolving why the visitor's wording was absent. I funded a destination model with a hospital-approved public name and approved aliases, used consistently by the kiosk, directory and print output. The hospital retained authority over what each destination represented. Our team did not create clinical equivalences or assume that similarly named services were interchangeable.
The programme cost ₹21 lakh over six weeks. It required product, content and site teams to agree ownership and review the affected outputs, which delayed a catalogue refresh. A customer delivery lead preferred local text overrides because they were faster to install. I rejected uncontrolled overrides where they would recreate the inconsistency elsewhere. We supported an approved correction process and made the relationship between the public label and aliases explicit.
In a later thirty-person exercise, twenty-seven matched the instruction to the intended destination and three still needed assistance. That was evidence about the tested naming problem, not proof that every language or destination was resolved. We kept those remaining difficulties in the research account and required site review for additional names. The company did not claim to have reduced clinical waiting time or independently verified all hospital content.
The outcome was a product obligation to carry a coherent destination identity through the journey. I still regard the hospital as the owner of its information, but ownership does not remove our responsibility to provide a structure that can keep it consistent. A person should not need to understand the provider's internal abbreviations before the navigation product can help them.
Interrupting a route when cached information could no longer support it
In 2025 we rehearsed eighty route-closure tasks using simulated site updates. Thirteen continued displaying an old route from cached information after the product could no longer confirm its status. The display looked complete and the route remained easy to follow. That apparent confidence was the problem: the interface did not communicate that the content needed to support the direction was unavailable or out of date.
The delivery team proposed retaining the last route until a new one arrived, because an empty map would generate support calls. I agreed that a blank failure message would be unhelpful, but continuing to direct a visitor was not an acceptable substitute. We defined an unconfirmed-route state that explained the limitation and offered a staffed-help option agreed with the hospital. Engineering implemented the state handling, and site owners remained responsible for approved routes and assistance arrangements. The product did not calculate an unapproved alternative or present itself as an emergency evacuation system.
The revision cost ₹32 lakh over nine weeks and displaced an additional visual customisation feature. We repeated the thirteen known stale-state tasks; each now interrupted the old direction and displayed the help route through the interface. This meant a way to seek assistance, not a new physical route through the building. We retained further testing for other failure conditions rather than extrapolate from those thirteen to every closure or network interruption.
The change created an operational discussion with customers about who would maintain the help information and what a visitor should do if the named assistance point was unavailable. Those responsibilities needed agreement before installation. I reported the extra configuration and support work as part of the product decision, not as an unexpected burden caused by engineering.
The exercise did not involve an actual emergency or establish a clinical benefit. It showed that a product could become more truthful by interrupting a seemingly smooth interaction. My lesson was to make uncertainty an authored part of the experience. A map that cannot currently support a direction should help the visitor recognise that limit before acting on it.