Making sense of prototypes and models

With several contemporary movements addressing faster, leaner innovation to choose from – including design thinking, agile engineering, lean startup, value proposition design and customer development – it’s sometimes difficult to separate similar jargon from different concepts. One such difference that’s confusing a lot of people is between prototypes and models and when it’s appropriate to use either in testing.

In this article, we make a clear distinction between prototypes as working products and models as simulated products. We first unpack the different uses of the terms today and second offer a simple guide to choosing the right method for testing a business or product concept.

Prototypes and models in the aerospace industry

The X-1, an experimental supersonic jet, is what many people think of first when prototypes are mentioned. Something about these daring pilots and cutting edge aerospace designs just capture our imaginations. It’s not surprising to find that dictionary definitions cite jet plane prototypes as use examples. A prototype in the aerospace industry is the first full-scale and usually functional form of a new type or design of a plane. To learn about necessary changes, engineers have to test the design under real flight conditions. In this use, prototypes occur only late in the development process.

But aerospace engineers test characteristics of the design in earlier phases too, using clay models. A clay model can be used in very early design phases to test the airflow over the fuselage and wings. While the distinction between a model and a prototype may be simple in the example of the aerospace industry, the language of modern innovation uses the words interchangeably and seems to treat the concepts as interchangeable too. This is not the case.

Prototypes and models in lean startup

At the core of the lean startup, is the minimal viable product (MVP). An MVP is a true prototype. It is a working product with the least features needed to satisfy a real customer. But very quickly new authors began to apply the term MVP to an ever expanding range of models such as web landing pages built to test the value proposition statement. MVP was soon applied to things that weren’t working products – prototypes – at all.  The models were very useful for testing aspects of the product before building a prototype. As tests methods go, these models have proven as useful to lean startup entrepreneurs as clay models (or computer simulations) are to aeroplane designers. Models help test non-product stuff – like advertising channels and value propositions – much earlier in the development process even before the availability of an MVP.

Web landing pages are a model category in lean startup collectively called smoke tests. Smoke tests include a phone number in an advertisement, a crowd-funding campaign, an online product demonstration video or a social media page. A solution interview is also considered a model because it exists to get feedback from customers on a proposed solution.

Around the same time that smoke tests were introduced as MVPs, terms such as mechanical Turk, wizard of Oz, and concierge were introduced. These are prototypes, not models. Wizard of Oz and mechanical Turk are synonyms for the same class of prototype – they are non-technical functioning service prototypes.

A great example of a mechanical Turk prototype is the online social question asking and answering service; Aardvark. When launched, Aardvark routed questions from users to other online users to answer the questions. To both users, it appeared to be an automated service but behind the scenes, Aardvark employees were doing everything by hand. They ‘faked it’ before building an automated product to replace human labour.

A wizard of Oz prototype, also called a concierge model, involves manually delivering the service to a single test case customer. A concierge prototype gives the red carpet treatment to a single customer or user to learn more about how the product or service meets the needs and wants of the customers.

Prototypes and models in design thinking

Long before the lean startup movement, design thinkers were building prototypes. In the early days when design thinking was still about physical objects, practitioners mocked up prototypes of products using materials at hand. Some prototypes were simple mockups while others actually worked. Apple’s original mouse is a trusty example of building the first prototype by cobbling together a butter dish, a deodorant roller ball and basic electronics.

As it became clear that smart design helped commercial products be successful, companies began using it in more contexts. High-tech firms that hired designers to work on hardware began asking them to create the look and feel of the user-interface software. Then designers were asked to help improve user experiences. Soon firms were treating corporate strategy making as an exercise in design. And today design is even applied to helping many stakeholders and organisations work better as a system.

You can see that somewhere along the way product prototypes got replaced with system models. But a common language was lacking. This common language Alexander Osterwalder supplied in Business Model Generation. I place the word model in italics to stress it. The business model canvas is a model of how the business does or could work. Later work in value proposition design introduced the prototyping of value propositions using the value proposition canvas. Nevertheless, keeping a strict definition this is an incorrect usage of the term. The value proposition canvas is a model, just like the business model canvas. It is not the real thing, it is a simulation on paper – it doesn’t deliver the product like a mechanical Turk or Wizard of Oz prototype does.

Models simulate some aspect of the thing while prototypes are working versions of the thing.

Prototype to Model Spectrum

When to use a prototype or a model?

In the end, does it make difference? As long as you understand what your goal is and the application of each – no it doesn’t. But if you’re not clear then yes, it makes a big difference.

Building a prototype – even a plain one – requires some technical expertise and resources. Building a model requires neither resources nor technical expertise. Building a prototype when a model will do wastes resources. Building a model when a prototype is needed will get you no insight.

So build a prototype only when you need to get insight into how customers really use your product. Build a model to test customer reactions to the product.