From Lemonade to Louis Vuitton: Building Better Business Models

In today’s fast-changing world, innovation is easy — but endurance is rare.
How do entrepreneurs design business models that stand the test of time? That’s the challenge our postgraduate students tackled in Entrepreneurship & Business Modelling, combining creativity, critical thinking, and AI-driven insights.

Read on to explore how they learned to turn imagination into impact.

• John Cleese's distinction between open and closed modes of thinking
• Sarasvathy's Effectuation Framework
• Why brainwriting produces better results than traditional brainstorming
• Mollick's concept of serendipity as "prepared curiosity” – The idea that entrepreneurial opportunities don't just happen randomly; they emerge when prepared minds notice possibilities others miss.

• What Large Language Models (LLMs) are
• How they process language through large-scale pattern recognition
• Where their strengths and limits lie across text, images, video, and document analysis.

Then I showed this image.

This formula turns AI from a search tool or simple content generation tool into what I call an intellectual augmentation partner.

Here’s one “perfect prompt” example we look at in class:

“You are a hospitality consultant specialising in customer-experience research. I’m exploring solutions for business travellers who struggle to be productive while staying at hotels in Bangkok. Identify five specific jobs-to-be-done that these travellers are trying to accomplish during their hotel stays. Present them as a numbered list using the format: ‘When [situation], I want to [action], so I can [outcome].’ Keep it practical and specific.”

Each element serves a function: persona frames expertise, context narrows focus, task defines the goal, format ensures usability, exemplar shows what “good” looks like, and tone keeps the result grounded.

Used this way, AI becomes less about generating answers and more about training clarity of thought. The better you think, the better AI thinks with you. Remember: you’re in the driver’s seat.

When AI Misses the Point

Students selected hospitality problems they cared about, then used the Perfect Prompt Formula to ask AI for five jobs-to-be-done. Most chose ChatGPT or Gemini.

The results revealed a familiar limitation we had just discussed in class: AI’s default mode is to please. It’s designed to sound confident, to agree, and to offer solutions quickly—even when the question calls for deeper reasoning. Students asked for jobs, but received solutions.

AI offered solutions such as:

“Business travelers need better WiFi.”

Instead of jobs-to-be-done such as:

“When I’m preparing for an important client meeting in my hotel room, I want to connect to reliable high-speed internet, so I can review large presentation files without technical anxiety.”

This exercise echoed some of the cautionary lessons we had just covered:

• AI lacks real understanding. It predicts plausible text; it doesn’t truly know.
• It’s sycophantic. Models are trained to be agreeable, which means they might validate a weak idea if you let them.
• You are in charge. The user—not the AI—bears responsibility for clarity, ethics and direction.

One student broke through. By progressively prompting—by insistently asking “why does this matter?” and “what outcome are they really seeking?”—the student guided the model beyond surface features to underlying jobs.

The exercise showed how easily AI follows the path of least resistance—and how quickly clarity returns when you take charge. This turned into an important lesson on agency when working with AI: if you want meaningful output, you have to lead the conversation, not let the model do it for you.

Working with the Business Model Canvas

From opportunity identification, we moved to business model architecture. I introduced the Business Model Canvas (BMC), Osterwalder and Pigneur’s framework for visualising how organisations create, deliver, and capture value through nine interconnected building blocks.

But why use a canvas at all? This connects directly to Steve Blank’s Lean Startup methodology, which underpins our approach throughout the course. Blank argues that startups aren’t smaller versions of large companies; they’re temporary organisations designed to search for repeatable, scalable business models.

The Canvas turns assumptions into hypotheses tested with real customers. Rather than serving as a static plan, the Business Model Canvas is a living system that evolves through validation.

This distinction matters because most entrepreneurial failures don’t stem from poor execution of good ideas; they come from executing ideas whose underlying business model was never tested. As Blank puts it, “No business plan survives first contact with customers.” The Canvas gives entrepreneurs something better than a plan. It gives them a testable framework. You can treat it almost like a test lab in which you strategically adjust how different pieces of a business model can fit together, refining and adjusting until you hit on a winning system.

From Lemonade to Louis Vuitton

With a fundamental understanding of the Business Model Canvas in hand, we then looked at business model examples across different levels of complexity.

A simple lemonade stand illustrated the basics: park visitors as customer segments, refreshing natural beverages as the value proposition, direct booth sales as channels, per-cup transactions as revenue streams, and ingredient and equipment costs as the foundation of the cost structure.

Then we scaled up. We examined Louis Vuitton’s multi-sided business model, which connects luxury consumers, brand devotees, and even investors interested in resale value. And we discussed Facebook, whose users function as both customers and products, with advertising revenue driving its network-effect ecosystem.

Our examples demonstrated how understanding different business model patterns helps entrepreneurs analyse their own industries with sharper eyes. In hospitality, a boutique hotel operates differently from an Airbnb host, which operates differently from a booking platform, which operates differently from a management company. We saw how the same industry can contain vastly different approaches to value creation, delivery, and capture.

• Which building blocks are interdependent?
• Where do the constraints lie?
• What would happen if you changed just one element?

Other assignments deepened both AI literacy and theoretical grounding: experimenting with the Perfect Prompt Formula on a small challenge, reading Steve Blank’s “Why the Lean Startup Changes Everything” in Harvard Business Review, reviewing *Business Model Generation*, and beginning to form project teams around new entrepreneurial ideas.

What’s Next: From Canvas to Customer

The Business Model Canvas is a crucial tool for aspiring—and active—entrepreneurs. Looking at the Canvas, you might easily think, “Aha! The plan for my new business is done!” However, as we’re learning in class, the Business Model Canvas, as powerful as it is, is only a foundation.

Next week, we’ll dive into the most critical relationship within that model—the link between Value Proposition and Customer Segment. Using the Value Proposition Canvas, students will map customer jobs, pains, and gains on one side, and products, pain relievers, and gain creators on the other.

This is where customer discovery begins. We’re creating hypotheses that must be tested in the real world.

The goal is to understand how everything connects: your value proposition must address specific customer jobs, relieve real pains, and create meaningful gains. If there's misalignment, no amount of clever marketing or efficient operations will save your business model. Get this relationship right, and the rest of the Canvas has something solid to build upon.

This is the entrepreneurial journey.You explore, test possibilities and learn new things about your business scenario. You iterate your way to a business model that works. Next week we’ll dive into how that learning starts with deeply understanding the customers you aim to serve.