How to Think, Build & Grow a Startup - Skills & AI Advantage | Varun Limaye | FO469 Raj Shamani

Summary

Varun Limaye argues that startup building can be taught in part. Founders cannot control luck or every market condition, but he estimates that 40–50% of the journey is controllable through learnable skills and repeated practice.

His central advice is to start with a real customer problem, test demand quickly, understand the economics, and build strong distribution—not to spend months polishing a solution before customers see it.

Startup and employment landscape

Limaye believes India will need more new ventures to create jobs, as large companies may not absorb the number of graduates entering the workforce. He predicts a shift from a few large employers hiring thousands (“10 × 1,000”) toward many smaller companies hiring smaller teams (“1,000 × 10”). He sees AI as accelerating this change while creating opportunities for small teams to become more productive.

He expects startups to compete with established companies on innovation, speed, and talent:

• Startups can run more experiments and have less legacy revenue to protect.
• Larger companies may hesitate to launch products that could cannibalize existing revenue, or move slowly because of hierarchy and brand risk.
• Some employees may prefer high-ownership roles at smaller companies over slower, more hierarchical organizations.

As examples of the “innovator’s dilemma,” Limaye recounts Kodak’s reluctance to shift away from film and Blockbuster’s failure to embrace Netflix’s online model.

Employment figures cited by Limaye are approximate claims from the conversation, not independently verified data. He estimates that around 1 million white-collar jobs are added annually, compared with 8–9 million undergraduate graduates. He says startups account for about 15% of those jobs, compared with 3% from government and 5% from manufacturing. He also cites unemployment rates of 3% for people without degrees, 30% for degree holders, and 40% for IIT graduates in 2024. The transcript provides no definitions or sources for these figures, so they should be treated cautiously.

Frameworks, processes, and playbooks

Start with the problem, not the product

• Identify a problem you understand, ideally one you have experienced yourself.
• Estimate its scale by considering how many people experience it, how frequently, and how intensely.
• Ask whether the problem matters enough that customers will pay to solve it.

Limaye’s point is that many startups fail not because of poor execution, but because they solve a problem customers do not care about. Think in terms of the customer’s job to be done. For example, some parents may use extracurricular activities to keep their children engaged outside the home—not necessarily to access an online learning platform.

Check the business model and unit economics

A business must both create value by solving a customer problem and capture value through sustainable economics. Before building, determine whether selling one unit can plausibly make money.

Break the business into its key cost and revenue drivers, then identify which levers the company can actually change.

Use a fast MVP and test demand

Limaye recommends getting a product or test into the market quickly—ideally within two weeks, where feasible—instead of spending a year building an unvalidated solution.

For a hypothetical product, he suggests creating a basic website, running a simple ad, and using a small test budget of around ₹5,000. Track:

• The ad’s click-through rate (CTR).
• The share of visitors who submit details, join a waitlist, or express purchase intent.
• Customer reactions to a proposed price.

Use the results to decide whether to improve, reposition, or abandon the idea. A crude early version can be more useful than a polished product launched too late.

Apply the “Mom Test” to customer conversations

• Ask about customers’ past behavior and real experiences, not hypothetical future intentions.
• Discuss their lives and problems rather than pitching a product and soliciting compliments.
• Talk less and listen more.
• Speak with people who have genuine stakes in the problem—not just friends or relatives who may be encouraging but are unlikely to buy.

Work backward from the information you need: Is the problem real and painful? What do customers use now? What is inadequate about those alternatives? Would they pay for a better solution?

Treat early customers as product partners

Find the right initial customer segment; the wrong users can give misleading feedback. Limaye suggests engaging an initial group—illustratively, the first 100 customers—in a direct feedback channel such as a WhatsApp group.

Keep releasing improvements and actively solicit feedback. Early customers can function as product managers, but the founder must interpret their input rather than blindly adopting every request.

Build a distribution strategy

In an “attention economy,” having a good product is not enough: a company must earn attention and make its message clear. Limaye recommends deliberately choosing:

1. Channel — where the audience can be reached.
2. Message — what the company wants them to understand.
3. Tone — how the message should sound to that audience.

He also recommends learning performance-marketing basics, including calls to action, targeting, funnels, creative testing, agency management, and relevant metrics.

Authentic participation on Reddit may help reach customers and influence how products appear in AI-generated answers. Overt promotion, however, can backfire in communities where users expect genuine contributions.

Use Amazon-style working backwards

Limaye describes Amazon’s PR/FAQ process: write the customer-facing press release and frequently asked questions before building or launching a product. This exercise requires a team to explain the customer value and experience clearly before committing resources.

He also describes Amazon leadership principles—especially customer obsession, bias for action, and “disagree and commit”—as embedded in interviews, performance discussions, and day-to-day decisions.

Learn by building, not only studying

Limaye says startup education should involve students building businesses, doing projects, and speaking with founders—not just learning concepts in the classroom. Repeated attempts build judgment and practical skills.

He argues that failure should be treated as feedback rather than a one-way door. The key is to keep testing ideas and learning from the results.

Example: Quick-commerce unit economics

Limaye illustrates how to analyze a business by breaking down a dark store’s economics:

• Store-level contribution: orders × margin, less rent, utilities, labor, and wastage.
• Order volume: potentially decomposed into the number of reachable customers, app penetration, and order frequency.
• Margin: influenced by average order value (AOV) and the platform’s take rate, less delivery and packaging costs.

Once less-flexible costs and constraints are understood, AOV and repeat ordering may be important levers for quick-commerce operators.

The same analysis helps explain why the model may be harder to extend to lower-density areas: fewer potential orders within a delivery radius may not cover the costs of operating a dark store. A technology change such as drone delivery could alter the delivery radius and customer base, while also introducing new costs or operational failures.

Example: Building a startup ecosystem

Limaye describes Stanford and Silicon Valley as an ecosystem rather than a university-only success story. He credits Fred Terman’s efforts to keep talent connected to Stanford, support ventures, develop a research park, and help early companies obtain customers and institutional backing.

He cites HP as an early example and says an early HP product was purchased by Disney. His takeaway for India is that universities, founders, customers, government, and institutions can reinforce one another by helping new ventures find talent, research, and initial demand.

Other business opportunity raised

Limaye identifies hiring as a major unresolved problem for his own company, Mesa. He describes the need as moving beyond large volumes of applications toward a small, well-curated group of strong candidates.

He suggests a potential service that could help employers reach hiring managers directly, screen candidates—possibly with AI—and present a shortlist rather than thousands of profiles. This is an opportunity he would pay to solve, not a proven business model.

Presenters and sources mentioned

Presenters: Varun Limaye (guest; co-founder of Mesa School of Business) and Raj Shamani (host).

Examples and sources referenced in the discussion: Amazon and Jeff Bezos (leadership principles and PR/FAQ); Kodak; Blockbuster and Netflix; Stanford, Fred Terman, HP, and Disney; Y Combinator; and The Mom Test.