Mihir Prabhu — Product, experience & business analysis

01

About

I joined BETSOL as a software engineer and worked out fairly quickly that I was a mediocre one. What I was better at was sitting with the problem — why users dropped off, why two systems disagreed, why a team was stuck.

So I moved sideways into design and research, then into product ownership, and more recently into business analysis. That path wasn't a plan; it was mostly my manager taking a chance on me earlier than I'd earned, and me trying not to waste it.

Today I work across product and analysis: turning vague, contested problems into something a team can actually build. I've been lucky to do that on things that mattered — an enterprise SaaS product, a registration experience used by a million people a year, and a compliance overhaul that touched an entire organisation.

I'm still early in a lot of this, and there's plenty I'm learning as I go.

Based in

Bengaluru, India

Currently

Senior Business Analyst, BETSOL

Education

B.Tech, Information Science

Open to

Product & AI-focused roles

02

Selected work

Customer experience2021 — 2022

A registration flow that 95.7% of people abandoned

Problem: Only 4.3% of people who reached the client's registration page finished it. Comparable platforms were converting around 38%. Nobody was sure why.
My role: Led discovery and acted as design consultant and day-to-day point of contact. The work was shared with a UX designer, a digital marketing specialist, and our senior director.
Approach: Interviewed employees, service partners, and candidates matching the target personas. Built a quantified journey map for every step, audited the analytics funnel stage by stage, and benchmarked three competitors.
Outcome: Registration conversions improved by roughly 80%, and the client saved an estimated $2.5M in ad spend. The engagement grew from one project to seven.

What I learned

The analytics told us where people left. Only the interviews told us why — and the real blocker wasn't the interface, it was trust. We were asking for a date of birth and a social security number before we'd given anyone a reason to believe us. I'd have found that months earlier if I'd talked to users before opening the dashboards.

Product ownership2022 — 2024

Taking an enterprise backup product from zero to a real subscription business

Problem: Zmanda had strong enterprise credentials but no self-serve motion. To reach smaller customers it needed a genuine SaaS product, not a licence with a web page.
My role: Product owner — roadmap, backlog, user stories, and agile ceremonies. I was one of two people on the product side, working with an engineering organisation of 100+ and three designers who did the heavy lifting on design.
Approach: Learned SAFe and the product on the job. Rather than chasing feature parity, we looked at where interest died and found it wasn't features — it was the gap between starting a trial and paying. So we rebuilt that path.
Outcome: The product went from $0 to $200K+ in revenue and held an NPS around 2× the industry average. Customers included NASA, Boeing, Yale, Georgia Tech, and Saudi Arabia's National Cybersecurity Authority.

What I learned

I was handed this before I knew what I was doing, and the first few months of backlog I wrote were not good. What helped was admitting that to the engineers early — once I stopped pretending to have answers, they started telling me what was actually hard, and prioritisation got much better.

Business analysis2024 — present

Rebuilding a company's processes around an FTC mandate

Problem: New FTC directives required a client to restructure IT and business processes across the whole organisation, on a fixed regulatory deadline, with no room to be approximately right.
My role: Sole business analyst across roughly 18 months — requirements, process mapping, and system documentation. Also currently leading analysis for their CRM migration from Dynamics 365 and a legacy SQL system to Salesforce.
Approach: Mapped how work actually happened versus how it was documented, then sequenced changes so teams weren't rebuilding and complying at once. Worked with every stakeholder tier, including the CEO, COO, and CTO.
Outcome: The organisation is compliant, and we used the mandate as cover to fix long-standing process problems that had never justified their own project. Separately, an audit of licences and idle infrastructure surfaced around $480K a year in avoidable spend.

What I learned

The regulation was the easy part. The hard part was that twelve teams each had a reasonable, incompatible view of how a process worked. Being the only analyst meant I over-documented everything — which felt excessive at the time and turned out to be the only reason the migration is tractable now.

03

How I tend to work

/

Talk to people before opening the dashboard

Quantitative data shows where something breaks. Conversations explain why. I've learned the hard way that doing these in the wrong order costs months.

/

Write it down, probably too much

Most disagreements I've seen weren't disagreements at all — people were describing different things with the same words. Documentation is unglamorous and usually the fix.

/

Say "I don't know" early

I've been thrown into domains I didn't understand more than once. Admitting the gap quickly gets me help; hiding it just delays the same conversation.

/

Prefer the boring lever

The highest-impact change is rarely the most interesting one. It's usually a form field, a missing explanation, or a handoff nobody owns.

Where my curiosity goes

Most of what I'm learning right now is about working alongside AI.

I use Claude and Gemini through most of my day — research, synthesising interviews, drafting requirements, building small automations and agentic workflows so the same task doesn't cost me the same effort twice.

Some of that has turned into shipped work: conversational-AI and voice-agent prototypes on CCaaS platforms, built for client pitches. And some of it has just taught me where these tools are unreliable, which has been almost as useful.

I don't think this makes me an AI expert. It's a practice I'm building, and I'd like to keep building it somewhere it matters.

04

Currently

Working on

A CRM migration from Dynamics 365 and a legacy SQL system to Salesforce — mapping how the old systems actually talk to each other so they can be decoupled safely.

Learning

Structured AI and agentic workflow courses, and getting more rigorous about evaluation — how you actually tell whether an AI-assisted process is better or just faster.

Trying to get better at

Saying no earlier. I take on more than I should, and the work that suffers is usually the work that mattered most.

05

Recognition

Q4 2021

Our senior director called me "the Swiss Army Knife of the marketing team" after I picked up design, video, and research work while most of the team was away.

Q1 2023

Recognised for the transition into product ownership and a PI planning session that went better than it had any right to, given how new I was to the role.

Q3 2025

An Above and Beyond award for work across product and operations.

Three Above and Beyond awards, all of which were really about teams I got to work with — I've been fortunate with colleagues and with managers who handed me things before I was ready.