How Zepto built India's fastest-growing grocery startup from a WhatsApp group

Executive overview

Most early grocery delivery players in India built supply chains first and hoped customers would follow. Zepto did the opposite — starting from a WhatsApp group during the pandemic, they talked to customers on doorsteps, did deliveries themselves, and let user behaviour drive every decision.

The result is 10-minute delivery across 50 cities, 1.5 million orders a day, and a $5B valuation — built by two founders who hadn't yet started college when they began.

Building customer-backward beats building supply-chain-backward — and the economics follow.

Starting from zero: WhatsApp, pandemic, and YC

• Aadit Palicha and co-founder KB started delivering groceries to neighbours in Mumbai during the first COVID wave, with nothing lined up
• The product began as a WhatsApp group; the first "customer" was an elderly woman alone down the road who started adding friends
• First app (Kirana Cart — "corner shop delivery") was built in ~72 hours, two months after the first delivery
• A comment-flood on a YC partner post landed them an 8-minute call with Jared Friedman, who suggested it could be a real business
• YC provided the framework for rigorous PMF testing: good verbal feedback is not enough — are people actually repeating?

Why 10 minutes isn't just a gimmick in India

• Indian grocery purchase frequency is 4x higher than in the US; most consumption happens within 4 km of home
• Household sizes are smaller, four-wheeler penetration is low, and people buy perishables frequently — bulk buying isn't viable
• Customers kept telling them at the door: "I'd rather just go to my fruits guy in the morning" — convenience framing missed the point entirely
• 10 minutes replicates the doorstep delivery culture (milk guy, vegetable vendor) that already dominates Indian retail
• Retention at 1-hour delivery was ~3–4% week-over-week; after commandeering a store and running it themselves, repeat rates jumped visibly

The dark store model and owning the full stack

• Controlling fulfilment was the only way to guarantee delivery SLAs consistently
• Owning the stack unlocked all four pillars of customer experience: speed, quality, selection, and price
• First dark store launched in Bandra (Mumbai) — a cardboard-cutout prototype — and quickly outgrew the rest of the city combined
• Now stocking 45,000–50,000 SKUs: from oranges to earphones to hoodies
• Zepto Cafe — first-party food (coffee, snacks) from 250 sq ft inside dark stores — reached 100,000 orders/day within roughly a year of launch

Surviving the 2022–23 capital crisis

• March 2022: capital markets collapsed; Zepto had 7x less cash than its two main competitors
• "Execute or die" became the operating philosophy — forced efficiency gains much earlier than most consumer internet companies achieve them
• The Silicon Valley Bank collapse hit during a fundraise that took 7–8 months to close — the only unicorn round in India that year
• By April 2024: ~$1B GMV, 70% of markets profitable, most mature markets at 4–5% EBITDA
• The forced efficiency turned into a durable advantage: better capital allocation and more growth per dollar invested

Building the ads and AI stack

• Ad revenue grew from $40M ARR to $200M ARR in one year
• Built the full performance ad stack in-house: relevance engine, attribution, campaign management, automated keyword suggestions — trained on LLaMA
• AI prioritisation framework: focus internal build effort on applications core to the business (ads, search, customer support, supply chain forecasting); let third-party agents handle non-core work (accounting, legal)
• Over 50% of customer support tickets now resolved dynamically by a generative chatbot trained on Zepto's own data — not rules-based
• Product expansion driven by search data: when users searched for coffee, they built Zepto Cafe; when they searched for lipstick, they added cosmetics

On building in India and the founder mindset

• Indian engineering talent is world-class and significantly less competitive to hire than in San Francisco
• The post-2022 funding reset created risk-aversion across the ecosystem — the ambition to swing for large outcomes is still emerging
• Zepto's long-term goal: build a benchmark internet company out of India, comparable to Amazon or Mercado Libre — on a 20–30 year horizon
• On work culture: high-execution environments work when the mission is genuinely compelling — people give their best when they believe in what they're building
• Advice to his 17-year-old self: build for the love of building, not for money or status — the goal should be the chance to wake up and build again tomorrow