Login

How to create a hypothesis for your Pretotyping tests

Pretotyping tests are the perfect way to gain valuable market research for products that aren’t built yet. This could be a new product you plan to bring to market and you are looking for it’s product-market-fit or an extension to an existing product line and you want to find out what the pricing should be or the features it should have.

The main benefit to using Pretotyping over other forms of market research is that it is fast to setup and gives you high quality and reliable behavioural insight. All of this BEFORE developing any physical product or MVP. The cons however are that you need to have a clear idea of what you want are testing for.

Pretotyping is not great at giving broad data sets, it is perfectly designed to give you accurate insight into one specific variable at a time. 

Have no fear though, we are here to help you setup your Pretotyping tests by first giving you a sense of direction.

This direction comes from your hypothesis, as is one of the most important aspects of setting up a quality Pretotyping test. It will help direct the research and give you measurable and meaningful results.

What is a hypothesis?

A hypothesis is an assumption based on insights that have been gathered through previous research. It is not a guess on what you think is going to happen in your upcoming research, but more a theory on what the results will be for your upcoming research based on previous data gathered.

The data gathered doesn’t have to be extensive to start building a hypothesis, it doesn’t even have to be primary research, you can use existing data sets if needed. What matters is that you aren’t going into Pretotyping tests guessing what will happen, you have an idea of what you expect, then Pretotyping can validate that.

There are 6 forms of hypothesis that you can create and each have a particular purpose when running market research and gathering consumer insights. Let’s first look at the statements that aren’t ideal for Pretotyping and why that is:

1. Alternative hypothesis

This type of hypothesis statement will outline what you believe the outcome of your testing will be, based on previous insights of course. There are 2 subcategories within this that give it a direction or not.

For example a directional hypothesis in Pretotyping would look like this:

Launching a blue version of this product will improve conversion rates

A non-directional hypothesis would look like this:

Launching a blue version of this product will influence the conversion rates

Alternative hypothesis statements aren’t particularly useful when it comes to Pretotyping. As mentioned earlier in this article, Pretotyping excels when testing for specific areas and data sets. Running with a hypothesis like this will only make it harder to analyse the results later on.

If you did decide to run your tests based on this type of hypothesis, you would be looking to use the directional version so it gives the test a better focus. The only way this works is because you have the direction of where you think results will go AND the dataset you think will be influenced.

2. Null hypothesis

The null type hypothesis is an opposite to the alternative hypothesis. You will be looking at what won’t happen, or more specifically how this test won’t change any of the outcomes defined in the variables.

For example a null hypothesis statement in Pretotyping would look like this:

Launching a blue version of this product will not change the conversion rate

This type of statement is similar in use to the alternative hypothesis, it lacks the required focus to be truly effective for Pretotyping and there would be better ways to build a hypothesis for your tests.

3. Simple hypothesis

Simple hypothesis are just that, simple. They have little depth to them and just outline the relationship between the variables.

For example a simple hypothesis in Pretotyping would look like this:

Releasing new product colours increase conversions

There is very little use for a simple hypothesis in Pretotyping, it would be better to use an alternative hypothesis over this type.

4. Complex hypothesis

Despite it’s name, it isn’t that complex to create a complex hypothesis. It is similar in style to the simple hypothesis but instead has more variables to relate.

For example a complex hypothesis in Pretotyping would look like this:

Releasing new product colours will increase conversions, decrease customer acquisition cost and increase the size of our target market.

This type of statement is again not great for Pretotyping for similar reasons to the simple hypothesis. But it is actually worse than that.

Because of the way Pretotyping works, you want to run each test with as specific a goal as possible, if you are trying to test and measure multiple metric changes at the same time, you are likely to develop a test that is very complex and so you can’t easily attribute the increase in conversions to a particular area of the test.

5. Empirical hypothesis

This hypothesis is created as theories are still being developed and validated. It includes enough detail to make it relevant but without being too off course when the final insights are gathered.

An example of an empirical hypothesis in Pretotytping would be:

Launching a blue version of our product will increase the conversion rate more than launching a yellow version

It is great to use this type when you don’t have much data to start with and are just getting traction with your insights. Within Pretotyping, it isn’t the best option but gives you a good starting point to run tests against.

The reason for this is that you have a direction that you are testing, i.e. is a blue product better than a yellow product, and a variable to measure that against, i.e. conversion rates.

We’ve looked at the hypothesis statements that aren’t ideal for Pretotyping and discussed why, so let’s look at the best type of statement you can use.

6. Statistical hypothesis

This type of hypothesis is based purely on data you have already collected or verified through other sources. It is a logic-based analysis where you research into a specific area and gather insights based on that sample group.

An example of statistical hypothesis in Pretotyping would be:

Launching a blue version of our product will increase the conversion rate from 5% to 20% in the age group of 16 - 24 year olds

As you can see, this is our most detailed hypothesis yet. It outlines what we are looking at, launching a blue product, how that is going to effect our variables, conversion rate, and in which group of people that will happen, 16 - 24.

This is the perfect type of hypothesis for Pretotyping and what you should be aiming for. The problem is getting the data for your hypothesis, so let’s talk through that next.

6b. XYZ hypothesis

Similar to a statistical hypothesis, this type is endorsed by the father of Pretotyping, Alberto Savoia so is perfect for this task.

It structures your hypothesis in a way that gives it a lot of focus easily, all you have to do is fill in the gaps. Here’s how is looks:

At least X% of Y will do Z

As you can see, all that is left is to fill in the XYZ of the hypothesis to have your final statement. An example of this would be:

At least 5% of supermarket shoppers will buy hand cream for €12 - 15

Learn how Bosch was able to increase the price from 179€ to 199€ with the help of Horizon

How to get data for your Pretotyping hypothesis

As shown the perfect hypothesis for Pretotyping is a statistical one, but to be able to create on of these you need to first gather data to inform your hypothesis.

Use surveys to gather broad insights to develop your Pretotyping hypothesis

The best way to gain initial data to inform your Pretotyping hypothesis then is to run more traditional market research methods like surveys. The great thing about gathering insights through surveys is that they can be broad in the data they gather.

In our example around launching a blue version of a product you could start by initially running a survey to capture insights around the current product and what respondents would think about different colours of product.

At the end of this round, they mind find that no-one wants a blue version and in fact would be more likely to buy a green version.

The problem here is that all of that data is biased because of the way it is collected, so if you were to develop a green version, when it came down to it, those same respondents might not even purchase it.

This is the reason you need Pretotyping, to validate that data. But you need that data to develop a valid Pretotyping test.

Now you have that data you can start to design your hypothesis and develop the tests you want to run to validate your hypothesis. If you have multiple points of data that need to be tested, don’t try and run these with one hypothesis, do multiple hypotheses for multiple tests to ensure your Pretotyping is focussed.

Can you use Pretotyping to get hypothesis data for Pretotyping?

Yes and no. You can use Pretotyping to gather initial market research data, but it will be across a lot of tests and you will start to lose the benefits of Pretotyping this way.

The way you would set this up is to setup a testing funnel. In essence, a lot of tests go in and insights come out that can help you develop a hypothesis.

Initially you would decide what you are looking to do with your product. Are you looking to launch a new colour as in our instances above? If so, you would need to start running multiple tests for target group and then features to get your initial results.

For the target group tests for example, you would start off testing colours to a target group of 16 - 24, then the exact same to 25 - 34, then the same to 34 - 42 and so on and so on. Then you would need to do the same for those groups, but with different interests and so on. 

All in all, it’s really a lot of work to gather the sufficient insight to create a hypothesis from Pretotyping data, if you are starting from nothing.

Sign-up for our monthly newsletter to get advice on how to predict market success and increase revenue with with robust purchase-intent data from real consumers, pre-market.

Hi! 👋 Who should we deliver this too?

How to create a clear hypothesis for Pretotyping

We have everything we need now to create a hypothesis, we have the statistical insights that are going to give us a point of reference and we know what kind of hypothesis we want to use, but there is still some work to be done to make a clear hypothesis.

It’s much like creating a clear product brief or task, communication is key so that everyone can understand the purpose of this test at a glance.

1. Address the problem

You have a problem or problems that need to be solved, that’s why you’re running these tests in the first place. This needs to be shown within the hypothesis statement, you need to clearly address what the problem is.

For example:

How does a new colour of our product affect the sales?

2. Keep it short

The hypothesis isn’t an entire brief into the research you are undertaking, it’s an outline statement of the research problem that you are trying to solve and what you think the outcome will be based on previous insights.

3. Define the variables

You need to understand what the factors for success are, if you are running a test, how will you measure the results and how will you define if the problem has been solved at the end?

You can put any variables you think are relevant here, it is best to keep them to the absolute minimum to keep the research focussed and attributable. 

Remember, if you need to do multiple tests to solve the problem or gain the insights you are looking for, that’s fine.

Variables could be conversion rate, it could be double-opt in, it could be cost-per-lead. You also want to add in your product variable, such as product colour or price.

4. Create final phrasing

Now you have all of the pieces for your hypothesis it’s time to put the final phrasing together that will direct all of the research underneath it.

Congratulations! You now have the best hypothesis possible to make sure your Pretotyping tests:

  • Have a clear goal
  • Can be attributed easily
  • Have a measure of success

Learn how Bosch was able to increase the price from 179€ to 199€ with the help of Horizon

Written by
Steven Titchener
An experienced growth marketer now helping Horizon and it's customers create successful products. Always looking to expand his ideas and take on unique and interesting takes on the world of marketing and product development processes.
LinkedIn Profile Link

More insights from Horizon

Want to learn to increase profits with validated price decisions?
Join our newsletter to learn how to identify price uplift opportunities with robust purchase intent data from real consumers, pre-market.