See. Decide. Act — in seconds.

Sprinklr's mobile app existed, but the supervisor experience was underdeveloped and unstrategic — features added ad hoc, no persona-grounded framework, no validated use-case hierarchy.

Supervisors were managing critical moments from WhatsApp, Teams, and phone calls — not because they wanted to, but because Sprinklr had no good mobile answer for them. Every minute of delay between a problem surfacing and a supervisor acting was a potential SLA breach.

The work sat squarely inside Project Saral — Sprinklr's executive-led simplification effort — so the pressure wasn't just to ship mobile features, but to ship the right ones.

I structured the study in phases — persona grounding, generative discovery, customer validation, and IA validation — each phase building on the last and answering a different part of the product decision.

The guiding constraint came from the product context: mobile real estate is scarce, supervisor attention is divided, and the moment a supervisor opens their phone is almost always a moment of urgency.

Every high-value mobile scenario followed the same arc: something changed, the supervisor needed to understand it in under ten seconds, form a judgment, and take one meaningful action — without opening anything else.

That became See → Decide → Act: the structural backbone the MVP could be built around. The supervisor isn't trying to do their job on mobile — they're trying to stay in control while temporarily away from the tool that lets them do it.

Six findings shaped the MVP. The most consequential:

The research didn't end with findings — it ended with a backlog. 9 epics, 40+ stories, telemetry definitions attached to each, phased across releases 26.4 and 26.7.

Epics mapped the See → Decide → Act arc: landing surfaces, alerts & interventions, then the system requirements (audit, RBAC, telemetry, accessibility) that made the rest trustworthy.

I treat MVP definition the way a PM should be able to: scope is a thesis, not a wishlist. So I sequenced the study to answer specific decisions, not to gather a buffet of insight.

Phase one was persona grounding across four customer sources so I wasn't reasoning from a single account's quirks. Phase two was generative discovery — open enough to surface what supervisors actually do when they're away from their desk, structured enough to ladder back to use cases. Phase three was customer validation on the emerging architecture, and phase four was IA validation on the queue-monitoring home snapshot specifically — because that's the surface that has to be right or nothing else matters.

I ran a V1 → V2 methodological upgrade on the survey mid-stream after the V1 instrument under-discriminated between supervisor and operations-manager intents. That's the kind of move I make in flight rather than waiting for a retro.

Three decisions shaped what the team built — and what we deliberately did not build:

Engineering got a Confluence space with the See → Decide → Act framework, 9 epics with linked user stories, telemetry definitions (event name, properties, success threshold) attached to each story, an IA spec for the home snapshot, and a residual-risk register naming what we knew was deferred and why.

PM got the prioritized release plan (26.4 then 26.7), a decision log of every scope cut with the supporting evidence, and a customer-validation summary tying each in-scope item back to the accounts that asked for it.