Bharat Kini

Work

Case studies that show how I think, decide, and execute.

Each case study follows the same shape: what the situation was, what I owned, what I did, and what changed. Outcomes over process.

Fintech Onboarding Mobile

Reducing drop-off in a high-friction onboarding flow

A KYC-heavy onboarding flow was losing 42% of users before they reached their first transaction. I led the redesign from discovery to launch, cutting drop-off by more than half.

Context

A fintech app offering investment accounts had a six-step onboarding flow required by regulation. Drop-off had been climbing for three quarters. Engineering had shipped incremental fixes without a structured diagnosis.

My Role

Product lead for the onboarding squad. I owned the problem framing, research synthesis, and the final spec. I worked closely with one designer and two mobile engineers.

What I Did

  • Ran six user interviews focused on the moments where users gave up
  • Mapped the funnel step-by-step with session recording analysis
  • Identified that two steps — document upload and liveness check — caused 61% of exits
  • Proposed progressive disclosure: surface the heavy steps later, after users had established momentum
  • Wrote the PRD, facilitated design review, and coordinated the A/B test setup

Tradeoffs

Reordering steps meant compliance review. I worked with the legal team to confirm the reorder was permissible and documented the rationale in the spec so it would survive team changes.

42% → 18% Drop-off rate at onboarding
+40% First-transaction completion
−28% Support tickets about onboarding
B2B SaaS AI tooling Content

Building an AI content brief tool for a content-heavy B2B platform

Content teams were spending 3.5 hours per brief in a fragmented workflow across docs, decks, and briefs. I scoped and shipped an AI-assisted brief generator that cut that to 45 minutes without compromising quality.

Context

A B2B platform serving content marketers had a significant gap between customer intent (fast, quality briefs) and reality (slow, inconsistent briefs spread across tools). Sales was flagging it as a retention risk.

My Role

I owned the feature from idea to GA. This included the AI integration spec, prompt design rationale, and the editorial quality rubric used to evaluate output before release.

What I Did

  • Interviewed eight content managers to map the real brief workflow
  • Scoped an MVP: a form-driven flow that feeds structured inputs to a language model
  • Wrote the AI output evaluation rubric (used by QA and editorial reviewers)
  • Defined the trust model: AI drafts, human edits, no auto-publishing
  • Prioritised the features that saved the most time without requiring AI to do the hard thinking

Tradeoffs

We deliberately constrained what the AI could generate — no auto-populate of strategy sections. This was slower to ship but earned trust with editorial leads and avoided the refund requests the previous beta had triggered.

3.5h → 45m Time to complete a content brief
91% Brief quality score (editorial rubric)
+22% Feature adoption at 60 days