I spent three years at ShopX. It was the most formative role of my career. Not because of the scale (that came later at Paytm) but because I built something from nothing and made every early mistake myself.
ShopX was a B2B marketplace connecting brands to distributors across India. When I joined, the platform had two links in the chain: brands shipped to distributors, and distributors managed everything downstream. Retailers weren't on the platform at all. They ordered inventory over phone calls or WhatsApp. My job was to add retailers as the third link. Build the channel. Get them on the app. Make them stay.
Zero retailers existed on the platform before this project. By the time I left, the channel had grown to 30K+.
Figuring out if the opportunity was real
I ran market research and competitive analysis before we committed to building anything. Reliance and DMart had started pitching retailers on digital tools. But their approach was narrow. Mostly accounting software. ShopX was planning something more complete: ordering, delivery tracking, inventory analytics, debit-credit ledger, and ticket management in one app. If we moved fast, we'd be first movers on the full-stack solution.
I built target clusters in priority order. Which geographies. Which retailer segments. Which distributor networks had enough density to support a new channel. Then I worked with engineering to customize the platform.
The two-app compromise
We planned to launch in 2 sprints. We hit that timeline.
One early compromise: distributors initially received two separate apps. One for ordering from brands. Another for receiving orders from retailers. The plan was a unified portal, but combining both flows added months of complexity. Tracking incoming and outgoing orders required different data views, notification logic, and dashboards.
I chose to ship two apps and merge later. We launched, validated that retailers would actually use the platform, and delivered the unified portal two months after. That decision shaped a principle I still follow: the imperfect version that ships teaches you more than the perfect version that doesn't. We built the unified portal with confidence because we had real usage data, not assumptions.
The onboarding problem
I expected the hard part to be building the product. It wasn't. The hard part was getting retailers to use it.
These were small shop owners in tier-2 and tier-3 towns. Most had run their inventory through phone calls for years. We weren't asking them to switch from one app to another. We were asking them to switch from no app to an app.
The activation threshold was 2-3 orders. Once a retailer placed a few orders and saw the delivery tracking, the ledger, and the analytics, they stayed. Before that, they'd default back to calling their distributor.
I designed the onboarding system in four layers:
Field playbooks. Consistent pitch, demo flow, and objection responses for the sales reps going door to door. The most common objection: "I already talk to my distributor on the phone, why do I need this?" The playbook reframed the answer around what the app tracked that a phone call couldn't.
In-app video kit. Thirty-second videos. How to place an order. How to track delivery. What to do if something goes wrong. Short, visual, minimal text. These users wouldn't sit through a tutorial.
1:1 setup sessions. A call between the retailer and someone from the in-office team. Not a support call. A trust call. Retailers needed to know there was a real person in an office connected to the person who showed up at their shop.
Bi-weekly check-ins. Every onboarded retailer got a check-in every two weeks to catch concerns before they became churn.
This system was labor-intensive. It wouldn't scale the way a self-serve flow would. But for this user base, self-serve would have failed. These retailers needed someone walking them through the first few orders. The product culture glorifies self-serve onboarding. For good reason, usually. But some users need a human to get through the first steps. The 1:1 sessions were the most "unscalable" part and probably the most important.
Today I'd layer a conversational AI agent into the onboarding flow. WhatsApp, which these retailers already used daily. Handle the common setup questions and objections automatically, escalate to a human when the query gets complex. The human touch would still exist, but for 70% of cases, the AI could get a retailer through their first order without a person on every call.
The wrong pitch
Six months in, adoption was growing but slower than it should have been. Retailers signed up but some never placed a first order. Others placed one and disappeared.
I visited retailers across 8 towns. Sat in their shops. Watched how they worked. Asked what they needed.
The answer was embarrassing.
Our messaging was wrong. All of it. ShopX had been a brand-to-distributor company for years. Every communication used distributor-oriented business language. "Grow your revenue by X." "Optimize your supply chain." "Scale your business with better distribution."
That language meant something to a distributor managing a large operation. It meant nothing to a shop owner in a tier-2 town drowning in messy inventory and missed deliveries.
Retailers didn't want to "grow their business." They wanted someone to fix the chaos. Tell me when my delivery is coming. Track what I owe and what I'm owed. Help me figure out what to reorder before I run out. Stop losing my orders.
We were solving their actual problem. But we were describing the solution in a language they didn't speak.
I redesigned all retailer-facing communications. Rewrote the pitch decks. Retrained every sales rep on a new message: we fix your operations, we cost less than your current process, and here's what we do that Reliance and DMart don't. The new pitch led with relief, not growth. With simplicity, not strategy.
The change drove organic inbound leads for the first time. Retailers started telling other retailers. Word of mouth from a shop owner who felt understood turned out to be the best acquisition channel.
With current AI tools, I'd approach this discovery differently. Instead of one field research cycle across 8 towns, I'd generate 20 messaging variants and A/B test them across clusters within weeks. Different value props, different lead messages, different tones. Let the data pick the winner. The field research would still matter for the "why," but the "what works" could be tested much faster.
Building the metrics layer
ShopX had no retailer-level analytics when I started. I stood up the tracking layer with no prior infrastructure. Daily dashboards covering adoption rate, reorder rate, order frequency, drop-off, delivery timeline, and NPS. Every lifecycle stage had visibility.
On-time delivery rate moved from 85% to 94% through iterative process changes. Each cycle, I reviewed where delays clustered and worked with ops to fix the bottleneck. Dozens of small fixes, not one big one.
RTO dropped from 10% to 1-2%. That was a team effort. The supply-chain team built predictive models. I surfaced the data patterns and coordinated with logistics partners. I wouldn't claim that reduction as mine alone.
What I took from this role
Your pitch is part of your product. The platform was good. The messaging was wrong. That gap cost months of slow adoption. Once we described ShopX in the retailer's language instead of the company's language, growth accelerated. I've seen this pattern since then. Teams build the right thing and describe it in the language of their internal culture. The fix is usually simple once you see it. Seeing it requires spending time with users.
Ship the imperfect version. The two-app compromise was ugly. It worked. We learned from real usage. Waiting for the unified portal would have burned months and taught us nothing.
Hand-holding is not a weakness. The 1:1 sessions and bi-weekly check-ins were the least scalable part of the onboarding system and the most effective. Sometimes you need a human to walk someone through the first few steps before the product can stand on its own.
This note covers the D2R (Distributor to Retailer) project. I also worked on D2C pilots with L'Oreal, Coca-Cola, and Unilever, and a restaurant delivery pilot during my time at ShopX. A separate note covers those projects.