Chief Product Officer / India
What the convenient workflow left for people to repair
Decisions & consequences
01Replacing a spoken-script estimate with a complete programme plan
In 2022 I approved a timing preview that estimated the duration of the main spoken script. For a thirty-minute planning window, the preview showed sixteen minutes used and fourteen available. Riniki challenged the account by walking through the complete planned sequence: sixteen minutes of main script, eleven of recorded clips, three of separate links and two of opening and closing material. Those components totalled thirty-two minutes. The calculation had answered a narrower question while its presentation implied that it described the programme.
The delivery sponsor proposed retaining the current preview and adding a note that other material might affect duration. That avoided rebuilding the planning view and left experienced producers to make the adjustment themselves. I rejected the proposal because the missing components were central to the task the feature claimed to support. My approval had helped create the misleading answer. We allocated ₹34 lakh to show the components together and make the two-minute excess visible against the thirty-minute window.
The revised product did not automatically remove two minutes of content. It gave the producer the evidence needed to revise the plan, including the assumptions behind the spoken-duration estimate. Actual delivery could still vary, and the workspace did not control transmission. The supported result was a more complete preparation account, not a claim that every subsequent programme would finish exactly on time or that a measured on-air failure rate had improved.
We gave up the simpler preview and had to make the timing model intelligible without overwhelming the producer. That required a different design discussion from merely changing the displayed number. I would now ask which components occupy the window before approving a spare-time claim, and I would test the answer against a whole rehearsal rather than one field. The mistake was not that sixteen plus fourteen failed to equal thirty. It was that the product had allowed a correct sum to stand for an incomplete programme.
Keeping production notes out of the presenter's reading copy
A 2023 export review used twenty-four synthetic programme packets. Nine included internal production notes in the copy intended for a presenter, while fifteen kept them separate. The notes were fictitious coordination material created for the test; the exercise did not expose actual guest details. It showed that our product treated a shared programme document as though every part served the same audience and purpose.
The workflow sponsor wanted to keep one export with a visual style distinguishing notes from spoken text. That was convenient for teams accustomed to working from a single packet and avoided another choice in the interface. Riniki argued that a different type style did not settle whether the material belonged in that output at all. I supported a ₹28 lakh redesign separating production notes from presenter copy, with a clear account of which fields each output carried and who should review it before use.
A follow-up synthetic test contained twenty packets. Eighteen separated correctly; two were flagged for correction rather than counted as successful exports. The release account retained those two exceptions. The test did not establish universal privacy compliance, security certification or error-free handling of all real programme material. Engineering and security owners had their own responsibilities, and producers retained the review of what they intended to use. The product finding concerned this defined separation task.
The change added a decision and some explanation where the old flow had offered a single convenient download. I considered that cost justified because the convenience had hidden a difference in purpose. We did not claim fewer actual broadcasts of private material or a measured reduction in incidents from a synthetic test. I would now begin an export design by naming the person and task it serves. A programme can have several legitimate working copies without asking every recipient to identify and disregard material that should not have reached their copy in the first place.
Letting a rehearsal edition stay still while the draft changes
In 2024 a twenty-four-producer study examined how teams used our collaborative draft during rehearsal. Sixteen wanted a stable reading edition, while eight preferred to see changes immediately. The existing product treated live propagation as the natural expression of collaboration. For the first group, that could move the text while somebody was working through it, leaving the team to coordinate the difference outside the workspace.
The collaboration sponsor preferred one live mode and suggested asking presenters to pause when a colleague edited. The proposal kept the product simple and preserved the immediate feedback that some teams valued. I thought it imposed the product's preferred rhythm on the other task. We funded ₹48 lakh to support a stable rehearsal edition alongside the continuing draft, with deliberate comparison and adoption of a revised edition. A live mode remained available for teams that wanted it.
In a later pilot of fifteen teams, eleven selected the stable edition and four selected live collaboration. Those were choices within a selected pilot, not evidence that the same proportions applied to all production teams. The work did not establish a retention increase or a broadcast-quality improvement. It showed that both modes could be represented in the product without pretending that a stable copy had stopped the underlying draft from evolving. The team still needed to know which edition it was using.
The decision required more explanation and a clearer transition between draft and rehearsal copy. It also made an unnoticed switch less acceptable: a new edition should not silently replace the one a presenter believes is in front of them. Producers retained editorial approval and programme decisions. I would now examine the moment at which a shared working object becomes something another person must rely on for a particular task. Collaboration can mean making change possible without requiring every participant to experience that change at the same instant.
Offering pronunciation help without replacing the original name
A 2025 preparation study involved thirty-six selected producers. Twenty-two wanted an audio cue when preparing an unfamiliar name, while fourteen preferred a written cue. The proposed feature had assumed a single phonetic-text field would be sufficient and had placed that field where the original name appeared in the reading view. That made an aid look like a replacement for the source text.
The design sponsor preferred the single written field because it was easy to search, copy and print. Those were useful properties, and the fourteen producers who preferred written cues gave us reason to retain them. I did not think they justified making the same form compulsory for everybody or removing the supplied name. We funded a ₹32 lakh prototype with optional audio and written aids displayed alongside the original, with the source and permitted use of an aid addressed through the relevant production process.
Thirty later preparation tasks used twenty-one audio aids and nine written aids. These were task choices, not thirty distinct people whose preference could be generalised to the market. Nor did choosing an aid prove accurate pronunciation. The prototype did not claim that a name has one pronunciation in every language or context, and it did not remove the preparer's responsibility to seek clarification when needed. No actual guest audio or identity is represented in this fictional account.
The product became more flexible and somewhat less compact. It needed an intelligible relationship between the original name and the optional help, rather than a field that quietly substituted one for the other. I would now ask whether an assistive feature preserves the material it is meant to help someone use. A cue can support confidence without becoming an authority it has not earned. The useful outcome was a choice of preparation aid with the underlying name still visible, not an unsupported claim that technology had solved pronunciation for every presenter.