Chief Product Officer / India
A compact result could still be the wrong result
- Product, research and design colleagues
- 28
- Three leads; ₹7.9 crore annual function expense
- Company revenue
- ₹112 crore
- FY 2025–26; 260 employees in March 2026
- Design-software experience
- 26 years
- Support through research and product leadership
Decisions & consequences
01Making the usable fabric boundary part of the layout decision
In 2022 a review of sixty customer-authorised layout examples found twenty-two that crossed the usable fabric boundary specified for the work. The product had asked for nominal roll width and used it as the available area. In one example, the roll was 152 cm wide but the customer excluded 1 cm on each side, leaving 150 cm usable. A proposal that fitted inside 152 cm could therefore still be unsuitable for that job. The numerical input was valid; the product had asked it to represent the wrong condition.
I had accepted the single-width interaction because it made setup quicker and the demonstration easier to follow. We paused the affected proposal path and worked with customer specialists to distinguish nominal width from the usable boundary they intended to apply. Product and design changed the input sequence and the visual explanation. Engineering implemented the revised boundary behaviour and checks. The user could see the excluded area and understand why a previously compact arrangement no longer fitted. We did not prescribe the exclusion for every fabric or claim that entering it replaced physical inspection.
The correction cost ₹23 lakh over six weeks. We reviewed the defined sixty examples again and documented the supported behaviour and remaining conditions. Some arrangements became longer or required a different placement because they could no longer use the excluded area. We retained that consequence rather than comparing their compactness with the earlier invalid result. The work did not establish a universal material-saving percentage or certify the resulting garments. It repaired the relationship between the customer's stated condition and the product's proposal.
The lesson for me was that reducing the number of inputs is not always reducing the user's effort. An omitted distinction can return later as a harder decision at the workbench. We changed our product review to ask which physical condition each collected dimension represents and how the user can inspect its effect. The choice belongs in the interface because it changes what the result means. It should not be left as a caveat that appears only after the customer has accepted a layout.
Withholding shorter suggestions that ignored a direction lock
A 2024 review of thirty-six layout suggestions found nine that did not preserve a customer's locked one-way orientation. The arrangement used less length on screen by turning certain pieces, even though the stated rule did not allow that change. The suggestion looked efficient only because it solved a less constrained problem than the user had asked. I had supported an earlier release account that emphasised compactness without making the status of those constraints sufficiently visible.
We stopped presenting the affected suggestions as usable outputs. Product, research and engineering worked with authorised customer examples to trace where the orientation restriction was lost and to decide how an unsupported result should appear. Six of the nine suggestions were repaired within the specified constraint. Three remained unavailable in the supported automatic path. We showed that limit and retained a manual planning route rather than removing the direction lock or labelling the unrestricted arrangement as an equivalent answer.
The correction and review cost ₹18 lakh and delayed the relevant release by five weeks. Where a valid arrangement was longer, we kept the longer result in the product demonstration. We did not compare it with the invalid arrangement to claim a realised material saving. The thirty-six examples formed a defined review set; they did not prove every direction rule or fabric property had been tested. Customers still needed to decide whether their specified conditions were appropriate for the actual work. The product's responsibility was to preserve the conditions it claimed to use.
The commercial discussion was difficult because the shortest layout made the strongest demonstration. I argued that a recommendation loses value if the user must rediscover which instruction it ignored. We changed the acceptance question from how tightly the pieces could be arranged to whether the proposal satisfied the constraints presented alongside it. That did not eliminate the need for further engineering or specialist review. It gave the team a more honest basis for deciding what could be released, what should remain manual and what the product should say it could not yet do.
Separating a neat page from a pattern at the intended scale
In 2025 we reviewed a print-preparation task with eighty participants using an authorised pattern example. A fit-to-page export made the drawing occupy the page neatly but changed a 100 mm reference to 97 mm in the reviewed output. Twenty-nine participants did not notice the scale change before treating the print as ready for their next task. The outline was complete and the export had succeeded, which made the problem less obvious than a missing piece or a visible error message.
I asked the team to separate output intended for discussion from preparation intended to preserve working scale. The redesign made the selected purpose explicit, showed when resizing was applied and retained a measurable reference. Product and research reviewed the explanation with users while engineering changed the export behaviour. We also made the need to check the actual printed reference clear. A software setting cannot establish that every printer, driver and downstream handling step will preserve dimensions; the user still needs an appropriate verification before physical work.
The redesign and review cost ₹31 lakh over ten weeks. The task findings described those eighty participants and that defined workflow, not a population-wide error rate. We did not claim that the revised interface certified production output or removed every possibility of resizing. Illustrative printing remained available for its legitimate purpose, with its scope distinguished from working-scale preparation. The change was about making the choice understandable and preventing a successful-export message from implying more than the product could establish.
I had approved the earlier convenience preset because it reduced setup and support questions. The review showed that fewer questions at export could mean a more serious question later. We changed product reviews to follow the output into the next task and ask what the receiving person would reasonably assume. The correction required extra interaction and a more careful explanation. I accepted that cost because the user needed to know whether the page was a picture of the pattern or a preparation intended to preserve its dimensions.