- The call
- Build B2C acquisition entirely on organic content infrastructure. Zero paid budget, at any point.
- The alternative
- Paid channels were the default marketplace playbook. PR and email acquisition were also tested properly, then deprioritised.
- The reasoning
- Course-marketplace unit economics don't absorb high CAC, and paid traffic stops when spend stops. Organic content compounds: built once, it keeps performing without proportional input.
- In hindsight
- The engine reached 200K+ MAU and kept producing traffic after the team's attention moved fully to enterprise. The one gap, and the part worth reading the full story for, is what we got right about email by accident the second time after getting it half-wrong the first.
Three case studies · 2019–2026
The
Work
This page is a body of evidence, not a gallery. It covers one platform built over four years across three distinct phases, preceded by two formative engagements in very different worlds: a multi-thousand-crore consumer electricals enterprise, and a healthcare lending product built from a proof of concept. Each case study shows the decisions, not just the outcomes, including what was deliberately not done and why.
Careervira · 2021–2026
The Role
Manager, Product & Product Marketing · 2021–2026. Joined at the MVP stage through an internship that converted to a full-time offer. Worked directly with the CEO throughout. Sole PMM for the enterprise motion, with cross-functional delivery across 16+ people in engineering, AI/ML, QA, and design.
Careervira began as an early-stage learning marketplace with strong supply and no demand discovery. Over four years it became a dual-engine business: a 200K-MAU B2C marketplace built entirely on organic acquisition, and an enterprise HRTech SaaS platform that grew to 10,000+ subscriptions and was rebuilt as an AI-native skill intelligence system. I worked across product and product marketing through all of it, from defining what to build to making it commercially viable.
Phase 1 · B2C scale
A programmatic content architecture across seven verticals, plus an early, decisive move into Google Web Stories, took the marketplace from flat growth to 200K+ MAU. 200K+ MAU · 20× in two years · ₹0 CAC.
Phase 2 · Enterprise buildout
Customer discovery with HR, L&D, and CHRO stakeholders shaped the product, the positioning, and a GTM motion built from scratch by a single operator. 10K+ subscriptions · 10%+ MoM growth · 5 personas designed for.
Phase 3 · AI-native transformation
The platform evolved from content delivery to skill intelligence. I owned the business layer: translating stakeholder workflows into AI/ML requirements. 10+ ML models · 75+ pipelines · role-based automation across 5 personas.
Crompton Greaves Consumer Electricals · 2021–2022
Operating-system design inside a large Indian enterprise
The Role
Summer intern, Trade Marketing (2021), returning as Management Trainee reporting to the Head of Global Sales (2022). The internship converted to a pre-placement offer. Both projects ran inside a business with 7 plants, 30+ distribution offices, and 10,000+ SKUs, which meant nothing shipped without coordination across sales, IT, and external vendors.
Project 1 · FOS Incentive Automation and Sales Mapping
The situation. Crompton wanted to introduce a performance-based quarterly incentive scheme for its distributor salesmen (FOS). The blocker was foundational: there was no reliable way to map secondary sales to individual salesmen at the distributor-retailer-division level. Doing it manually, every quarter, across thousands of salesmen, would have made the scheme unworkable and politically risky. An incentive programme that pays the wrong people, or pays late, is worse than no programme at all.
What I built. I acted as product and project manager for a three-stage feature built into Crompton's Distributor Management System. Stage one: FOS identity architecture, a master upload model using distributor code and Aadhaar to generate a unique, persistent ID for every salesman, with a replacement workflow to handle FOS turnover. Stage two: retailer-division mapping, connecting each FOS to the retailers they actually served. Stage three: reporting integration, including a Tally integration so distributors could see FOS attribution directly on their invoices. The delivery work was as real as the design work: I prepared the Scope of Work, coordinated requirements between sales, IT, and the external vendor, and led the data collection, cleaning, and validation across the FOS base.
The outcome. FOS data compiled for 1,700+ distributor salesmen at 97.5% coverage. Over 100,000 retailers mapped across distributor-division combinations. The Trade Marketing team could disburse incentives centrally, without depending on the field sales organisation, which removed the bottleneck that had made the scheme unviable.
- The call
- Treat this as a product design problem, and build identity architecture into the DMS, rather than treating it as a data-collection exercise.
- The alternative
- The expected approach was a collection drive: have the sales team gather FOS and retailer data into spreadsheets, reconcile centrally, and compute incentives from the result. Faster to start, and it would have produced a usable first quarter.
- The reasoning
- Incentives recur every quarter, and FOS turnover is constant. A spreadsheet snapshot decays the day it's completed; by quarter two you are re-collecting everything and re-fighting every dispute. A persistent ID architecture with a replacement workflow makes the mapping self-maintaining, and putting attribution on the distributor's own invoice through Tally made the data visible where disputes actually happen. The extra build time bought a system instead of a snapshot.
- In hindsight
- Held up. The 97.5% coverage figure is really a product of the architecture: unique IDs made gaps findable, and findable gaps got closed. The transferable lesson is one I have reused since: when a process has to repeat, the deliverable is the system that runs it, not the output of the first run.
Project 2 · Centralised Tele-Sales Pilot and SOP Design
Crompton wanted to test whether centralised tele-sales could improve dealer and retailer outreach, standardise follow-ups across regions, and reduce field-team dependence for routine order collection. Before any national rollout, the model needed ground validation. I led the Bangalore pilot end to end under the Head of Global Sales: running outreach with dealers and retailers, observing response patterns and operational frictions, documenting the workflow, and translating field learnings into a Standard Operating Procedure. That SOP became the basis for Crompton's national centralised tele-sales programme. The project's significance is straightforward pilot-to-scale thinking: validate on the ground, codify what works, and hand the organisation something it can run without you.
Milaap Social Ventures · 2019–2020
0→1 healthcare lending, first hire on the initiative
The Role
First hire on the healthcare lending initiative, inside Strategic Initiatives, a four-person unit, then Senior Associate, Business Development (April 2019 – March 2020). A first job in a high-trust, high-stakes environment with no playbook: the work ran across hospitals, NBFCs, patients, and internal teams simultaneously.
Project 1 · Healthcare Lending Product, POC to Operations
The situation. Milaap, known primarily for crowdfunding, was building a lending product to address a specific failure mode in Indian healthcare access: patients who needed treatment but lacked immediate funds, and who often didn't qualify for traditional financing because of credit-score thresholds, documentation gaps, or sheer treatment urgency. Hospitals were losing patients to affordability-driven drop-offs.
What I built. I joined at the proof-of-concept stage and worked alongside the project head on the product's early design and its operating model. The model ran across two product flows (Milaap-funded loans and partner-NBFC loans) and two transaction types: Instant Loans for time-sensitive cases such as dental clinics and emergency care, with eligibility decisions inside 30 minutes, and Planned Loans requiring deeper credit and prognosis evaluation, on a 12 to 24-hour turnaround. I coordinated across hospitals, NBFCs, and internal teams, supported provider onboarding, helped operationalise eligibility logic across multiple financiers, and contributed to the revenue model: commission from lenders plus service fees from hospitals, with the hospital fee often passed through as a patient discount to enable interest-free financing.
The outcome. ₹1 Cr+ disbursed within two quarters of operations. An 88.5% reduction in turnaround time, achieved through successive SOP refinement as the operation matured. An early-stage idea became a structured, multi-stakeholder financing operating model.
Project 2 · Crowdfunding Lead-Channel Development
Milaap held unutilised funds across older inactive campaigns, in many cases where the original beneficiary had passed away and families had asked for the remaining funds to be redirected to other patients. Existing lead channels weren't surfacing enough live, time-sensitive treatment cases to deploy those funds responsibly. I built the alternate pipeline on the ground: partnerships with Tier 1 hospitals in Karnataka, systematic activation of an existing base of 400+ medical partners across South, East, and North India, and direct engagement with senior stakeholders at fintech firms and NBFCs to identify lending partners. Incoming cases were qualified by treatment urgency and stage, routed to the crowdfunding teams, and tracked through the cycle. In parallel, I ran informal audits of defaulters on the lending side to flag potential NPAs and shape mitigation. The result was a ground-led pipeline of urgent treatment cases that let dormant campaign funds reach the patients they were meant for.
- The call
- Build the product, the partner network, and the operating process in parallel, rather than sequencing them.
- The alternative
- The textbook sequence: stabilise the product and eligibility logic first, then build the hospital and NBFC network, then scale operations on top of a settled process.
- The reasoning
- At the POC stage, with a four-person unit and patients whose treatment couldn't wait for our roadmap, the three workstreams were the same workstream. You could not design eligibility logic without live cases, could not get live cases without hospital partners, and could not keep hospital partners without an operation that actually disbursed. Parallel building was the only honest way to learn what the product needed to be.
- In hindsight
- The call was right for the stage, but it had a real cost: process refinement lagged behind volume, and the 88.5% turnaround improvement came from SOP work that, in retrospect, could have started a full quarter earlier. What I would sequence differently now is not the workstreams but the codification: write the SOP from case one and treat every early transaction as an input to it, instead of letting the process live in people's heads until the volume forces it onto paper. That habit, codify while you operate, is one this project taught me permanently.
You've just read the whole body of work: four organisations, five years, and the reasoning behind every major call. If the way this work was decided is the way you want work decided around you, let's talk.