Chief Product Officer / India
Restoring the page, keeping the choice, preparing the chapter
- Product organisation
- 21 colleagues
- Three leads across reader and publisher workflows
- Annual function expense
- ₹5.6 crore
- Includes ₹1.9 crore discovery and design
- Design and product career
- 27 years
- Visual studio support through illustrated-story product leadership
Decisions & consequences
01Restoring the words my automatic crop had removed
In 2019 I approved an automatic page crop intended to make illustrated stories easier to follow on a phone. The adapted view filled the screen with a cleaner sequence of pictures. In a selected review of twenty-four pages, nine lost part of a speech balloon while fifteen remained intact. The affected words changed what a reader could understand. The attractive visual presentation had concealed a failure to preserve the story.
The design team wanted to refine the crop because it made better use of the small screen. Nazia and the participating creators argued that the product needed a dependable way to show the original page and inspect adaptations before release. I accepted that the problem was not simply a setting to make the automatic view more aggressive or more conservative. We needed the workflow to preserve the creator’s ability to examine the result and the reader’s ability to return to the page.
We spent ₹29 lakh over seven weeks restoring a full-page fallback and adding creator preview before adapted views were released. All nine affected pages in the selected review were corrected. The fifteen initially intact pages remained part of the reviewed set; we did not convert that result into a claim that the whole catalogue had been examined. Publishers and creators retained their respective approvals, and the product did not declare a general accessibility certification from the correction.
The trade-off was retaining a less uniform visual experience and adding a review step instead of presenting the automatic crop as ready for every story. The bounded result was that the selected pages preserved their words and the workflow made adaptation visible before release. No measured improvement in completion or sales is claimed. I kept my approval error in the case because the cropped preview had appealed to my own design judgement. The decision taught me to ask what a cleaner interface has removed from the work it promises to make easier to experience.
Shipping a choice between guided panels and the whole page
In 2023 we explored how readers wanted to move through illustrated pages. Eighteen adult participants took part in the selected study. Ten preferred guided panels, and eight preferred seeing the whole page. The product team could have treated the larger group as a reason to make guidance the single default experience. I thought that would turn an exploratory observation into a much broader claim about the way everyone should read.
Nazia’s research showed that the whole-page view helped some participants understand composition and anticipate relationships between panels. Others valued help moving through a dense page without having to navigate each detail themselves. Creators also wanted to inspect how a guided sequence related to the work they had made. We chose to preserve both experiences rather than make one disappear because it was slightly less common in a small study.
We used ₹46 lakh over twelve weeks to release optional guidance with a page overview for seventeen pilot titles after creator review. Readers could use the guide and return to the original page context. The overview was available within the product; this case does not claim a certification of accessibility or an evaluation of every assistive technology. The title set and participant group were limited, and their results were not presented as a market-wide preference percentage.
The trade-off was greater design and implementation complexity than a single enforced reading mode. We accepted that cost because the study had made two legitimate uses visible. The bounded outcome was a released choice in the pilot titles, not an asserted increase in retention, completion or revenue. Further work could examine how people switch views across different stories and over longer reading sessions. I wanted the product decision to reflect the useful evidence we had without borrowing certainty from a sample too small and specific to settle every reader’s preference.
Allowing the next chapter to be prepared without the whole volume
In 2025 our offline preparation flow required a complete volume before the reader could treat any of it as ready. That suited some people, but was excessive for someone who only wanted the next chapter. The selected fictional example contained six chapter payloads of 20 MB each, total 120 MB. A person planning a short reading session had to prepare the whole 120 MB content payload under the existing product flow.
The engineering team favoured the simplicity of one complete-title state. Research showed the usefulness of a smaller, deliberate preparation choice, and publisher colleagues wanted the reader to understand exactly which content was available offline. I supported a chapter-level option while retaining the complete-title choice. The decision was about the unit of content transferred and the clarity of its status, not a change to the publisher’s permission or a claim that the files had been compressed more effectively.
We spent ₹54 lakh over fourteen weeks developing the workflow. In the selected example, a reader could choose one 20 MB chapter instead of the mandatory 120 MB volume payload. The other five chapters remained visibly unavailable offline until separately prepared. A completed chapter state did not imply that the entire work had been stored. Publisher and rights requirements remained part of the existing product responsibilities, and readers who preferred to prepare the full title could still do so.
The result was a more granular choice about what to transfer before reading away from a connection. We did not claim a universal download time, network improvement or payload size across the catalogue. Nor did we infer a measured increase in reading completion from the feature’s availability. The trade-off was a more detailed content-state model instead of one simple whole-book status. I accepted that complexity because it let the product describe the reader’s actual preparation more accurately and made the next reading intention possible without requiring a larger transfer the person had not chosen.