As the co-founder of LinkedIn, Reid Hoffman once said that trying to innovate and scale a business quickly requires you to "throw yourself off a cliff and assemble your airplane on the way down." Sounds risky. It doesn't have to be. 

For most new upstarts -- be it a startup or corporate venture -- finding the perfect product-market fit is critical before you run out of money. Rapid experimentation helps you reduce risk and increase your chances of success. 

Having worked as an innovation consultant for 25 years, and being the co-founder of my own SaaS company, I've experienced the challenges involved in defining, testing, refining, and scaling new ideas. Every startup founder feels the pressure to reach critical mass as quickly as possible. Big companies face similar challenges, mostly because internal funding comes and goes with annual budgets. 

One success factor applies to anyone creating a new venture: The need to identify and test assumptions underlying the offering and business model. Most innovators now know the language of minimal viable products (MVPs), A/B testing, agile and lean. While we may understand these concepts philosophically, it's important to use practical tools to rigorously define and track experiments over time.  

When Zappos set out to sell shoes online during the early days of the internet, it needed to test a single big assumption: that people will actually buy shoes online. The hypothesis was that people would click and purchase. Its experiment involved creating an online store using photos from a local shoe retailer. When the company sold a shoe, it would buy the shoes at retail and mail them out. Simple. The approach validated their model. They grew the company and eventually sold to Amazon.

An "experiment canvas" lays out the key assumptions and risk factors related to launching new products and services. The canvas is essentially a one-page document that literally helps your team get on the same page about what you're doing and how to make sure you're doing the right things. 

So how do you create an experiment canvas? Just list your riskiest assumption, the hypotheses related to it, and the experiments you'll do to test your hypotheses. Creating your template is the easy part. Getting your team to agree on the content can be tough, but a hugely valuable part of the process.

Here's what to do: 

  1. List your riskiest assumption: What is the riskiest assumption about our offering and the business model we follow? (Zappos needed to test whether or not people would actually buy shoes on the internet.)
  2. Define hypotheses: What hypotheses must we test to validate or invalidate our assumption? (Zappos tested whether people would click and buy shoes without trying them on.)
  3. Conduct experiments: What experiments will test these hypotheses? (Zappos built an e-commerce website with photos of shoes it didn't own.)
  4. Review data: Did the data validate or invalidate the hypotheses? (Zappos had enough people buy shoes that it was confident it had validated its hypothesis.)
  5. Make conclusions: Based on what we found, what changes should we make to our offering or business model? (Zappos went for it and scaled the business.)

It's important to determine what kind of data needs to be collected as an appropriate metric to decide if the hypothesis is refuted or supported. When the experiment is finished, you should have enough data to draw a meaningful conclusion. Your conclusions should focus on whether or not a change in direction is needed. The entire process can and should be repeated over and over until you've run out of risky assumptions. For many founders and innovators, it's a never-ending process.