arrow
Back to blog

Business Ideas Validation: 5 Best Experiments for Hypothesis Checking

clock

10 min read

When we create an idea, it’s similar to falling in love. You see no flaws in your creation; there are no downsides to what you’re passionate about, and this is absolutely normal to feel that way. However, the target audience does not always share the creators’ point of view, causing many startups to fail.

Usually, when we create something, our view is biased by our own experiences and desires. On the other hand, people are ultimately pickier about the products or services you offer since your competitors already offer them similar things. So, as soon as you have an idea, you enter the playground where you have to prove your concept to be top-notch not for yourself but for your audience. And there is no better way to see how your product will do on the market than to try and experiment.

Business idea validation is what this article is about. Here I will explain the hypotheses notion, how they help build your product, and how to experiment with minimal risks and maximum efficiency. Because your idea in a vacuum can be the absolute best, but as soon as you hit the market, you notice that everyone thinks their idea is no less impressive. Therefore, we’ll explore how to clear your vision based on my article and “Testing Business Ideas” book by David J. Bland and Alex Osterwalder.

Why 20% of Ideas Fail

Usually, more than two thirds of startups will never succeed on the market. Some sources say that over 20% of businesses end up faltering in the first year of their existence. And, the most dire reason for failure is no market need with nearly 35% of startups declining because of it. On the graph below, you can see other causes of startups failure.

Top reasons why startups fail

Most of these failures incentivize startups to be able to leverage the competition and hold on to the market with proper customer analysis and business hypotheses for promotion. For example, timing, market fit, cost issues, and even flawed business models can be changed with a trial and error approach.

So, in the following paragraph we will see more of it in action. Why do you need to hypothesize your customers’ behavior and how to check if your assumptions are correct.

What is Hypothesis in Business

A hypothesis is a supposition for a phenomenon when little evidence is present. Hypotheses are characterized by testing possibilities. So, for an assumption to be legitimate you need to perform the process of testing multiple hypotheses to validate it.

Hypotheses Types

  1. Feasible – the ease of realization. This type of hypothesis assesses the risk of your business to fail due to poor management, scaling, lack of resources (technology, brand, etc.), or partners.
  2. Desirable – the desirability of your idea. This type of hypothesis assesses the risk of not being able to retain the customers, acquire new audiences, or that too few people want your products or services initially.
  3. Viable – the potential of success. This type of hypothesis assesses the risk of your business not being able to generate enough revenue.

What differentiates a good hypothesis

A good hypothesis differs by several criteria. In this paragraph, I will unravel most of them as well as good hypothesis examples from my point of view:

  1. Testing ability. The hypothesis testing is needed to validate your assumption, otherwise there will be no use for it further.
    Example:
    Non-testable: Gen-Z users prefer making bank appointments online.
    Testable: Given a choice between calling or making an online bank appointment, we believe Gen-Z prefers the latter.
    How to make your hypothesis testable: try to use more definitive drafting to help you find the basis for further validation. Also, the best choice would be to use “if-that” formulation. In the non testable option above you see rather a statement, whether the testable option gives you more space to verify the assumption.
  2. Precision. Precise hypothesis is an assumption that also includes success metrics in the statement you make.
    Example:
    Non-precise: Young users won’t spend a lot of time reading our company’s blog.
    Precise: We believe users aged from 18 to 30 will spend on average in between 2 to 5 minutes reading an article.
    How to make your hypothesis precise: make sure you include all variables into the equation. Find out who you are making assumptions about, what are their average characteristics, and actions you’re trying to access. This way you will identify a success indicator and validate whether your business is capable of achieving it.
  3. Discreteness. Speak as clear and as narrow as possible. Keep your hypotheses focused on one thing and be as concise as possible with your targets.
    Example:
    Non-discrete: We believe consumers will use our software to make online bank appointments and check on the status of their financial operations.
    Discrete:
    - We believe users will make online appointments via our software.
    - We believe users will be interested in checking the status of their financial operations in our app.
    How to make your hypothesis precise: Separate your goals and identifiable indicators in order to get more concrete information when validating your hypotheses. Don’t try to check everything at once, it’s not an all-in-one solution. Give yourself time and space for experiments packed with evidence.
DashDevs can help you prove your hypotheses!
Find out how to achieve better market fit and accelerate time to market!

What to Do With Hypotheses

So, what to do with hypotheses? First of all, it takes time to learn how to make them right. Here’s an interesting exercise that can help you figure out how to prioritize hypotheses, take down the risks, and learn the cutting edge of assumptions mapping.

Start with having your core and supporting team. The former are people dedicated to the campaign’s success. Core team consists of people with honed design, technology, product skills or other competences depending on what you’re working on. The latter are people not directly engaged in the business endeavor you’re working on but essential for your success, like marketing, safety, or user research departments.

Then, you and your teams are going to undergo three similar steps in order to map the assumptions. Overall, this technique can help you improve the quality of your product market fit, customer centricity, research, and audience understanding.

  1. Start with hypothesis identification. Describe the hypothesis on sticky notes and put them on the canvas. Use different colors for each type of hypothesis.
  2. Prioritize hypothesis. Draw the graph on your hypothesis map; axis X is the evidence, axis Y is the importance. Then, proceed to locate your hypotheses on the map accordingly. The hypotheses you should test first are the ones without evidence and of highest importance. The assumption map
  3. Assess risks. Maintain focus on the top right quadrant of your assumption map since that is where the most risks lay. If you use unvalidated hypotheses with little evidence and they are false, it can break your idea down.

How to Test Hypotheses

The best way to test hypotheses is an experiment. It allows you to see how your theories might work in practice and sometimes these observations can save your business from failing. Before you start experimenting, however, let’s figure out the algorithm to follow.

The first step is to design an experiment. In the beginning, it’s better to go slow starting with small, cheap, yet effective experiments. The next step is to actually run an experiment. Be considerate of the time it takes to gather concrete evidence and data. Mind that for a sufficient survey you will need to question around 50 people.

What is Experiment in Customer Centricity Context

Experiments in customer centricity as well as in business development context are the procedures designed to help you reduce risks and obtain intel about your customers. It’s your best way to navigate through the market with your business idea, check its viability and demand customers have for your product.

However, what differentiates a good experiment from the pointless one is precision. You need to have three factors on hand: test subject, test context, test elements. Below you can see test cards that can help you figure out exactly how a good experiment looks like.

Test cards for experiments

First, in the “we believe that” blank you see the hypothesis under testing. Then goest the description of the experiment you’re planning to run in the “to verify that, we will” section. Next is a metric you will assess, evaluate, or calculate to define whether the experiment was in fact successful.
And lastly, is the criteria of success measured with the said metrics.

The Best Experiments for Hypothesis Validation: Top 5 by DashDevs

Now, finally, let’s get to a more practical question. I figure you’re interested in trying these experiments for yourself so in this paragraph I chose some of my favorites. In DashDevs’ workflow I use some of them on a regular basis, and for now they proved to be efficient.

So, without further ado, these are the best experiments for hypothesis validation from my standpoint.

Customer Interview

A customer interview test is one of the best ways to get quality insights on your customers’ pains and needs. This is a research experiment, it gathers solid evidence with 80% accuracy.

Customer interview is not a costly test and it doesn’t take a lot of time, so you can and you should conduct it as soon as you start developing your business. The average run-time for this experiment is about 30 minutes as the interviews don’t usually take a lot of time.

How to run:

  • Preparation: write a script for an interview including the major questions about customers’ desires or pains. Look for interviewees and select a proper time frame for data analysis.
  • Execution: ask questions according to the previously created interview script. Don’t forget to take notes, focusing both on answers and body language of your interviewees. Conduct at least 15 interviews.
  • Analysis: discuss the interviews, sort the notes, perform a ranking analysis, and update your value proposition according to the results.

A Day in The Life

A day in the life experiment is an observation test that allows you to dive deeper into the everyday routine of your customers. It generates solid evidence, is relatively cheap, and won’t take much of your time. So, this experiment is also startup-friendly.

How to run:

  • Preparation: define where and how your team will observe, and clear a few hours to dedicate the time needed for the experiment.
  • Permission: make sure those you’d like to observe give you full consent. Explain the reason behind your experiment when asking for permission.
  • Observe: capture the customers’ activity, jobs, pains, needs, and make notes. Don’t interact with your observants until the end of the observation.
  • Analyze: sort the notes, make takes, and update your value proposition accordingly.

Web Traffic Analysis

This is the experiment we at DashDevs are using, too. It is extremely insightful as it helps you to view how much time people spend on your website, how often they drop off, and the amount of attention your website gets.

How to run:

  • Preparation: define your focus area; for example it can be the number of downloads from your website or increasing sign ups. Select a time frame for your analysis.
  • Execution: use software to track web traffic. Not the drop off rates, and other indicators you’re interested in.
  • Analysis: check with the criteria within your focus area. How many sign ups are you getting? Is this the desired number? What can you do to improve the number?

Pinocchio Experiment / Pretend to Own

It’s time to use your imagination with this experiment! Create a low fidelity prototype, non-function but of a resemblance of the final product you’re working on. Then, pretend that this prototype is working exactly as you plan your final product to work and check whether this product actually fits in your and your customers’ routine.
You can do similar with MVP and paper prototypes.

How to run:

  • Preparation: sketch your idea, create the prototype, and estimate the time needed for an experiment. Don’t forget to maintain the log.
  • Execution: use your prototype as if it was fully functional and track your actions via the work log.
  • Analysis: analyze your work log to find cumbersome parts of your product. Use the log further to improve the final product.

Wizard of Oz

You remember the infamous tale about the Wizard of Oz? The experiment in question is really similar to what the said wizard was doing in the story.
In this experiment, you define the automation features that are yet too expensive for your business to adopt. Then, you ask the responsible people to do these tasks you wanted to automate, while pretending they are actually automated. This will help you assess whether those features are needed for customers’ satisfaction, will they help you generate more revenue, how many requests your team can perform manually, and at what volume you will need to automate your processes.
This experiment is the best way to prevent premature scaling.

  • Preparation: plan your workflow, create a tracking board for orders, test your production process and its effectiveness.
  • Execution: document how long it takes for you to complete each order. Gather satisfaction feedback from your customers.
  • Analysis: review your customers’ feedback and your worklog. Evaluate the delays you faced in the process of product making and how those delays affected your profit generation. Then, use this intel to assess whether it will be the right choice to automate.

General Overview

Overall, experiments are the best way to figure out how your product and company works without disrupting the processes. They can help you avoid most of the stumbling blocks in your business growth process.

So, I believe you can get maximum efficiency with the data you gather via the experiments. And, if you decide to make assumptions without proper testing and idea validation, this can do the opposite to your business.

Your view isn’t always unbiased, no matter how hard you try. You may want to deliver what you see fit, or what you think your customers see fit, when in reality people might want an entirely different thing. Don’t give people what you think they want. Give them what they say they want.

Share article

Table of contents