Chief Product Officer / India
The release changed when the promise was examined
- Engineering and product experience
- 28 years
- Career begins July 1998
- Product, design and research colleagues
- 38
- Led through four leads
- Annual function expense
- ₹9.4 crore
- ₹2.1 crore approved product investment
Decisions & consequences
01Reserving the room before calling the appointment confirmed
We rehearsed sixty appointments that required a practitioner and at least one additional resource. Nine appeared available in the product even though the required room or support resource had not been reserved. The practitioner calendar was doing what it had been designed to do, but the product's confirmation implied a broader commitment than that calendar could support. I had previously favoured the simpler release because it removed steps from the booking path.
Engineering proposed retaining the practitioner-first design and sending an exception list to clinic staff. That would have protected the planned release date, with manual coordination covering the remaining constraints. Research and clinic operations argued that the visible confirmation would still mislead the person making the appointment. I accepted that criticism and approved ₹74 lakh of work on a combined resource reservation. We displaced a planned reporting enhancement and delayed the release while the rules and failure states were examined.
The repeat rehearsal covered the same sixty scenarios and produced no recurrence of the nine known resource clashes. That is a defined test result, not evidence that every possible clinic configuration is correct. We retained the need for provider-specific setup and exception review. Clinical teams remained responsible for deciding what resources their services required and whether a particular appointment was appropriate; the scheduling software did not acquire that judgement.
The most useful change was in the product question. Instead of asking how few steps could lead to a confirmation, we asked what had to be true before the word 'confirmed' was justified. The answer included work beyond the screen. For a board considering a service-software investment, I would ask whether management has identified the obligations inside its product promise and the cost of supporting them. A simpler interface may still be the right answer, but the simplicity should follow an understood operating model rather than transfer invisible work to the customer after the transaction appears complete.
Keeping an earlier-slot offer from silently cancelling an existing appointment
A waitlist feature offered earlier appointments when a slot became available. In a forty-eight-person exercise, fourteen participants believed receiving the offer meant their existing appointment had already changed. The product used similar visual language for the offered and confirmed states. That ambiguity mattered even without any clinical judgement: a person could act on a time that had not been accepted or assume the original commitment no longer existed.
The growth team preferred a single-tap replacement with a short expiry message, arguing that additional confirmation would reduce take-up. Research wanted a more explicit account of which appointment remained valid. I chose the latter and approved ₹38 lakh for a redesign that separated the offer, acceptance and expiry. The original booking remained in place until the replacement was explicitly accepted. We accepted an additional interaction and postponed a promotional feature to make room for the work.
In the repeat exercise, forty-three of forty-eight participants distinguished the states correctly and five still required assistance. We kept those five in the result and in the support plan. The exercise did not establish a change in clinical waiting time or show that the flow would be understood by every future user. The product retained staff assistance and a clear account of the appointment that remained valid when an offer expired.
I learned that a state transition is also a statement about responsibility. An offer asks someone to decide; a confirmation records a decision already made. Treating them as interchangeable can make the conversion report look efficient while leaving the user's obligation uncertain. Directors need not design the screen, but they can ask what management counts as a completed transaction and whether that definition matches the commitment a customer understands. The additional step was worthwhile because it made the promise explicit, not because more steps are inherently better design.
Funding a keyboard route after the preferred interaction excluded users
Thirty keyboard users attempted a defined appointment-change task in a moderated exercise. Twelve could not complete it using the available controls. The preferred pointer-based interaction had performed well in earlier demonstrations, and our release account had given that path too much weight. The task itself was administrative: choosing a permitted change and understanding its effect on the current booking, without entering clinical details.
The delivery team proposed documenting a staff-assisted workaround and revisiting the controls in a later release. Research argued that the product should provide a workable direct route for the tested task. I approved ₹46 lakh for an alternative keyboard-operable path and associated review, moving a cosmetic dashboard refresh out of the release. We retained assistance for people who needed it, but did not use that assistance to declare the direct task complete.
Twenty-seven of thirty participants completed the task in the repeat exercise; three still needed help. We recorded the remaining difficulties instead of presenting the revised route as universally accessible. The exercise did not certify the whole product, cover every assistive technology or establish compliance with every applicable requirement. Further assessment and support remained necessary. My responsibility was to fund and prioritise the product work while allowing the researchers to report the result without pressure to remove the unresolved cases.
The decision changed how I discuss release readiness. A successful preferred path is evidence about that path, not permission to ignore other ways people must use the service. I would ask a board whether management knows who was excluded from its completion measure and what resources are allocated to improving their experience. The question should lead to a practical plan and honest limits, rather than a broad claim that exceeds the testing. The product became more usable for the observed group because we treated their task as part of the main offer, with remaining work still visible.
Making twenty-two unacknowledged changes visible to the receiving team
We examined 150 staff-made schedule changes and found twenty-two without an acknowledged handoff to the affected clinic team. The edit had been saved and the product displayed a completed status. That status described a technical event, but it did not show whether the people responsible for the revised arrangement knew what had changed. We needed to distinguish storing the change from transferring the work it created.
Operations proposed a daily digest so teams could catch up without adding another action to each edit. Product research showed that some consequential changes needed a named receiving role before the next working step. I approved ₹59 lakh for a workflow that made assignment and acknowledgement explicit, with follow-up when the expected recipient was unavailable. We accepted additional staff interaction and displaced part of the planned dashboard work rather than rely entirely on a digest read at an uncertain time.
Eighteen of the twenty-two known handoffs were closed after review; four remained under follow-up. The result does not claim that every revised appointment took place or that a clinical outcome improved. It records a narrower administrative responsibility. We left the four open cases visible instead of treating the software release as proof that all historical exceptions had been resolved. Staff still needed to examine those individual circumstances through their normal operating processes.
My conclusion was that completion belongs to the task, not merely to the save operation. A product can truthfully report that a record changed while still showing that a handoff is outstanding. For directors, the useful challenge is whether management's performance measures distinguish those states. Acknowledgement can itself become an empty ritual if nobody checks what it means, so the design also requires observation and feedback. The choice was to expose the unfinished work clearly enough for an accountable team to act, accepting the inconvenience of a status that remained open.