Abhiroop Kumar

Last week, Ashwini Vaishnaw silenced doubters at the World Economic Forum. Not with hype. Not with predictions. But with clarity. He said something most people missed: India isn’t just using AI. India is building all 5 layers of the AI stack. 1. Application 2. Model 3. Chip 4. Infrastructure 5. Energy This isn’t how most countries approach AI. Most pick one layer and go all in. India is building the entire foundation. He also made one thing very clear: 👉 The biggest ROI in AI won’t come from chasing the largest models.  👉 It will come from AI applications that solve real business problems at scale. That line stuck with me. So I went deep into what India is actually building at each layer: • Where we’re catching up to (and in some cases leapfrogging) the US & China • Why India could become the largest supplier of AI services globally • And which companies are already building across the stack No buzzwords. No future promises. Just what’s happening on the ground. I broke everything down in this carousel. If you care about India’s AI future — save this. If you’re building in AI — repost this. 🇮🇳 Follow Krishang Sheth for more content like this! #India #BuildInIndia

10

1 Comment

Last year, Marmik Mankodi and Nachammai Savithiri R wrote what is, in effect, a field manual for building a consumer-tech business in India. Today the reality is this: Starting a brand has become very easy. But getting it in front of customers has never been more expensive. If you look at the P&L of modern D2C companies, the line between “building a product” and “running a marketing engine” is fading fast. In some categories, 60–80% of revenue goes right back into performance marketing. What’s left is often a thin 5% EBITDA business pretending to be a high-growth brand. So the real question becomes: How do you build something that actually lasts? In this week’s episode, i sit down with Marmik Mankodi to break down what it really takes to build a consumer-tech business in India: A few things that we discussed , Why white-labeling is the safest way to test before spending big. How to calculate the real ROI of influencer and paid campaigns. Why the funnel has collapsed to “Acquire → Retain” If you’re a founder in this space or simply curious about how consumer businesses really work , this episode is for you. Link to the episode & the report in comments .

116

2 Comments

𝐂𝐨𝐮𝐥𝐝 𝐍𝐞𝐭𝐟𝐥𝐢𝐱 𝐢𝐧𝐭𝐫𝐨𝐝𝐮𝐜𝐞 𝐚𝐝𝐬 𝐚𝐧𝐝 𝐝𝐫𝐨𝐩 𝐢𝐭𝐬 𝐦𝐨𝐛𝐢𝐥𝐞 𝐩𝐥𝐚𝐧 𝐟𝐫𝐨𝐦 ₹149 𝐭𝐨 ₹99 𝐭𝐨 𝐚𝐜𝐜𝐞𝐥𝐞𝐫𝐚𝐭𝐞 𝐬𝐮𝐛𝐬𝐜𝐫𝐢𝐛𝐞𝐫 𝐠𝐫𝐨𝐰𝐭𝐡? Netflix has been struggling to grow meaningfully in India. With an estimated 12–15 million subscribers, it sits miles behind JioHotstar’s 300 million+ user base, powered by free content and cricket. Let’s run a simple, directional model. A price cut from ₹149 → ₹99 means Netflix must recover ₹50 through ads. Even if Netflix charges a premium ad rate of ₹200 per 1,000 impressions, it earns ₹0.20 per ad impression. To earn back the ₹50 discount per user - 250 impressions per user per month. Break this down: 👉 Assume average Indian mobile Netflix user watches ~1 hour/day 👉 250 impressions/month ≈ 8–9 ads/day 👉 ~4 𝐦𝐢𝐧𝐮𝐭𝐞𝐬 𝐨𝐟 𝐚𝐝𝐬 𝐢𝐧 𝐚 60-𝐦𝐢𝐧𝐮𝐭𝐞 𝐬𝐞𝐬𝐬𝐢𝐨𝐧 This already feels borderline—but Indian audiences are used to more ads on Youtube. Math is only half the problem—the strategy is the other half. 1️⃣ 𝐂𝐚𝐧𝐧𝐢𝐛𝐚𝐥𝐢𝐳𝐚𝐭𝐢𝐨𝐧 𝐫𝐢𝐬𝐤 A significant part of Netflix India’s base is already on the ₹149 mobile plan. A ₹99 ad tier would cause mass downgrades, shrinking revenue instead of expanding it. 2️⃣ 𝐈𝐧𝐝𝐢𝐚 𝐡𝐚𝐬 𝐬𝐭𝐫𝐮𝐜𝐭𝐮𝐫𝐚𝐥𝐥𝐲 𝐥𝐨𝐰 𝐂𝐏𝐌𝐬 Even a “premium” ₹200 CPM is far below the $45–55 CPM Netflix earns in global ad markets. 3️⃣ 𝐋𝐨𝐰𝐞𝐫 𝐮𝐬𝐚𝐠𝐞 → 𝐭𝐨𝐨 𝐥𝐢𝐭𝐭𝐥𝐞 𝐚𝐝 𝐢𝐧𝐯𝐞𝐧𝐭𝐨𝐫𝐲 If real watch time is lower than assumed, the ad-tier business collapses. 4️⃣ 𝐇𝐢𝐠𝐡 𝐜𝐨𝐧𝐭𝐞𝐧𝐭 𝐜𝐨𝐬𝐭𝐬 Expensive originals + low ARPU = weak margins. But here’s the overlooked final question: 𝐖𝐨𝐮𝐥𝐝 𝐚𝐝𝐯𝐞𝐫𝐭𝐢𝐬𝐞𝐫𝐬 𝐞𝐯𝐞𝐧 𝐰𝐚𝐧𝐭 𝐭𝐡𝐢𝐬 𝐮𝐬𝐞𝐫 𝐛𝐚𝐬𝐞? India’s digital advertisers prioritize cheap reach, not premium audiences. A niche, low-scale Netflix ad tier would sit 𝐥𝐨𝐰 𝐨𝐧 𝐚 𝐦𝐞𝐝𝐢𝐚 𝐩𝐥𝐚𝐧𝐧𝐞𝐫’𝐬 𝐩𝐫𝐢𝐨𝐫𝐢𝐭𝐲 𝐥𝐢𝐬𝐭—far behind YouTube, JioHotstar or Meta. And for Netflix, ads aren’t just a feature—they require building 𝐝𝐚𝐭𝐚 𝐢𝐧𝐟𝐫𝐚𝐬𝐭𝐫𝐮𝐜𝐭𝐮𝐫𝐞, 𝐨𝐩𝐞𝐫𝐚𝐭𝐢𝐨𝐧𝐬, 𝐦𝐨𝐝𝐞𝐫𝐚𝐭𝐢𝐨𝐧, 𝐜𝐨𝐦𝐩𝐥𝐢𝐚𝐧𝐜𝐞, 𝐚𝐧𝐝 𝐬𝐚𝐥𝐞𝐬 𝐭𝐞𝐚𝐦𝐬 𝐢𝐧 𝐈𝐧𝐝𝐢𝐚. That’s a massive fixed cost. Even if users tolerate ads, the economics don’t. #Pricing #Strategy

8

2 Comments

Netflix’s biggest growth lever in India costs exactly ₹0 in new content. While diving into the data for my latest Product Case Study, I found a massive blind spot. South India has huge mobile usage and streaming penetration. Yet, engagement with Netflix's massive library of Hindi originals in Tier-2 and Tier-3 markets was disproportionately low. It wasn't a quality issue. It was a language barrier. I realized that for Telugu, Tamil, and Kannada speakers, it didn't matter how good the recommendation algorithm was. If the subtitles weren't there, users weren't dropping off mid-episode; they just weren't hitting "Play" to begin with. To bridge this gap without inflating the production budget, I designed a scalable, AI-powered subtitle pipeline: 1️⃣ ASR + LLM Translation: Automating the heavy lifting of regional translations at scale. 2️⃣ Human-in-the-Loop QA: Maintaining the premium viewing experience Netflix is known for. 3️⃣ A/B Experimentation: Validating the impact on the North Star Metric, "Watch Hours per User", before a full rollout. This project was a great reminder: sometimes the most profitable product bets aren't about building the next big feature. They are about removing the invisible friction from the value you already have. I've attached my full slide deck below breaking down the problem and architecture. I'd love to hear from other PMs. What is the most impactful "low-hanging fruit" you've ever discovered in a product? Netflix What do you think about this solution? #ProductManagement #Netflix #GrowthStrategy #TechCaseStudy #ArtificialIntelligence

9

1 Comment

Fractal Analytics, India's first AI unicorn, had a muted IPO debut, closing below its issue price. This signals persistent investor caution around AI stocks despite India's push to become a global AI hub. The company also downsized its offering significantly pre-launch. #FractalAnalytics #IPO #ArtificialIntelligence #IndiaTech #markets

6

Sub-100 people running one of India's largest financial platforms — while managing ₹1,00,000+ crore in AUM. Nithin Kamath's 16-year update isn't a highlight reel. It's an admission: revenue hasn't grown in real terms, MTF risk keeps him up at night, and market share numbers he's told the world not to trust. That honesty is the actual moat. Everyone else posts growth screenshots. Zerodha posts the scary chart - ₹9,000 crore MTF book, 25% of net worth - and explains exactly why it's scary. Founders don't lose trust by admitting risk. They lose it by pretending there isn't any. 16 years in, Zerodha is still small on purpose. Maybe that's the whole point. #Zerodha #NithinKamath #StartupIndia #Fintech #FounderLessons #BuildInPublic

3

Last few years have been about Learning & Unlearning this beast called META. Common Mistake most D2C brands make - Treating Meta like a Performance Channel instead of a Discovery Engine. When we started scaling ALYV, we spent a lot of time figuring out how Meta actually works. A few learnings stood out: | Think Meta -> Discovery Engine Especially early on, use it to understand who responds, which problem resonates, what messaging, what CVP gets attention That learning is often more valuable than the first month’s ROAS. | Start broad. Let Meta find buyers The natural instinct is to keep adding interests and narrowing audiences. But giving Meta enough room to learn and not micromanage works better. To start, let Meta's Advantage Plus suite work its magic | Creative as a targeting lever. With ALYV, UGC consistently outperformed brand creatives, followed by video and then static. A “problem - solution” UGC will naturally attract a different customer than a “premium lifestyle” creative. Don't just make 5 versions of the same ad. Instead test different customer problems, product benefits, personas, social proof etc The biggest learning often comes from why something worked. | Ads Funnel Strategy For High ASP categories, this strategy usually works Top of Funnel - Creatives for Awareness Building, Attention Capturing, Education Middle of Funnel - Use Social Proof, Testimonials Bottom of Funnel - Offer Callouts, Discounts etc to close the deal | Creative fatigue is real – Give Meta more creative ammo Keep looking for actual deterioration in performance. Sometimes an ad could keep working; sometimes fatigue showed up much earlier. Expect better outcomes from creative diversification - multiple formats,  placements etc | CAPI + Pixel + Event tracking hygiene matters Better signals give the platform better inputs to optimize leading to better customer quality | Harvesting existing intent For an early-stage brand, allocate 15-20% of the budget for retargeting campaigns. For a mature brand, take it up to 40-45% | Don't ask Meta to do all the heavy lifting As we scaled ALYV, we realised that Meta can capture demand—but organic, referrals, retention and brand building help in creating demand. And once Meta starts plateauing, the answer isn’t always “Fix Meta” or “Spend More” Sometimes the answer is: Fix the Message or the Product or the Offer or the Experience Fix your fundamentals … Meta can only do so much to accelerate a good D2C business

6