©2019 by Lutav | Sérgio Tavares, ph.D.


A simple canvas to create an effective and aligned hypothesis for your design task.

A hypothesis canvas has a very simple structure. But in order to be effective for a business — that is, to save resources, accelerate goals and affect customer behaviours — it needs to include a few elements that are not part of the "scientific hypothesis".


These elements, included here, will save you time and ensure your design task will yield a business-oriented result. It doesn't matter if you are designing a new app idea, a change on your website, or a new business-to-business solution. This structure helps you to ensure you are keeping up with design thinking, and keeping your project human-centric, business-oriented and data-friendly.

In other words, these are your North Star audience, or those you can / will retain for longer time

When starting with your target audience, my usual hunch is to start with your VERY BEST CUSTOMERS (if you deal with an existing customer base). Why? Because you have reliable data on their behaviour, not only a hunch and demographics. Some designers like to start with a blank page. I think that is crazy. No business contacts a consultant with zero knowledge of to whom they will take the next step. Previous knowledge, however, should not be a legacy weight, but historical data for analysis. 


Either way, have in mind always the most loyal people.

State the ACTION you will execute in order to trigger a reaction (by incentive).

Could this question be simpler? Hardly. So give also a simple answer.

This helps the work to be precise and measurable.


Instead of "IF we make our site more appealing", state "IF we change our website to a one-pager booking system".


It's objective and goes straight to business objectives and goals (learn the difference between these two here).

And instead of dealing with long processes of "best scenarios", "worst case scenarios", use the structure of the hypothesis to plan and prepare for outcomes.

What's in it for my customer? Refrain from including what's in it for you, here.

Use this common sensical question to answer clearly and unmistakably: what does my customer get out of it?


If the benefit is unclear or hard to measure, you may come to the conclusion that it is NOT a problem worth solving.


Go back to your business analytics (number of customers affected, probability it will affect business) and rethink.

Lastly, benefit from scientific rigour to plan a null hypothesis, that is, if everything fails, what happens? 

You will be able to start thinking of the worst case scenarios, and adapt your solution to avoid them.  It will also help you to plan new experiments in case your null hypothesis prevails.


State also your desirable outcome (the alternative hypothesis): what if it works? You will be able to prepare clearly on accounting the outcomes, and improve the solution even if it works.

desired outcomes also can improved later on, on a next round of experiments.

Lastly, establish properly the time of your experiment to begin and to end. You may want to have enough time and subjects tested in order to reach statistical relevance. In business, however, decision are also made with most-likely-to-work solutions. Either way, manage expectations by setting the schedule for your hypothesis to be tested. Be generous. It should take the time it takes.


You may create your own canvas, but this is a working model I have used with many businesses, from small to big. 

Enjoy, and share if it was helpful!

P.S.: I don't have a newsletter. But I share a lot on LinkedIn. Connect with me there