“So, Tom. Can you explain what your business hypothesis was?”
“No,” came the curt reply.
“Okay, so let me understand this. You’ve got an increase in conversion rates — which, don’t get me wrong, is good — but you're not sure why. Is that right?”
“I made several changes last week, so it must have been one of them...”
“Yes, but which one?” I questioned.
Tom looked flustered as he searched for a plausible story. “The truth is Tom, you don’t know and that isn’t helping us increase conversion.”
I’ve lost count of the times I’ve had conversations like this.
.webp)
Tom was like so many other entrepreneurs who like to tinker.
Often driven by limited evidence, people like Tom make changes based on supposition. Their starting point comes from what they read online. If survivorship bias had a smell, entrepreneurial homes would reek like a field being fertilized with manure.
What they lack is a theory—a rational argument for why they are making a change. A business hypothesis explains that if you make a change, then this should be the outcome.
Unlike Tom, when you work from a hypothesis, you have a proposed explanation of what you expect to happen. You use good experiments to test your theory, giving you a conclusion that could stand up in court (if needed).
In this article, I will explain how I helped Tom incorporate a hypothesis-led approach into his way of working. You will have the foundations to begin using hypotheses as a framework in your work.
Let’s dig in.
What is a Hypothesis?
A hypothesis is a statement of your expectation of making a change. It's taking what you think will happen and framing it as an experiment to record the result.
As the Economist, Milton Friedman said:
The only relevant test of the validity of a hypothesis is comparison of prediction with experience.
Now a business hypothesis goes a stage further.
Through good experiments, It captures the expected value that the change will create. It enshrines the return on investment that demonstrates the opportunity on offer.
The ROI also creates an objective measure, giving a baseline to measure the outcome.
So, in summary, a hypothesis is:
- A hypothesis is a testable statement that reflects your predictions.
- It offers a theory of what might happen based on our experiences and knowledge.
- It is testable so that you can measure the outcome against your prediction.
Not Hypothesising: A Wasted Opportunity
Tom was trying to experiment, but not in a data-driven way.
When I began consulting with Tom, he gave me a long list of different things he had tried. He wanted to increase conversion so he changed the button users click on. Tom had tried different button sizes, various shapes, colours, and the copy — the list was lengthy.
But he had no idea whether they worked or not.
He had this pretty line graph with peaks and troughs. He tried to argue these were due to the changes he had made. But without a starting hypothesis, I shot down his argument.
“You’re not making fact-based decisions”, I told Tom. “ Even worse, you’re declaring success subjectively when you lack supporting evidence to back you up. You are wasting the opportunities you have infront of you.”
In truth, Tom was aflected by survivirship bias, confirmation bias, and outcome bias. Although he wasn’t taking a wild guess, Tom was taking insights from some ‘expert’ and using that to guide what he should change and when.
It’s a common sight among those who fail to be data-driven.
Without a hypothesis, Tom was defeating any hope of progress other than success via a fluke.
I remembered what Gavin, an old accountant friend of mine once said:
Businesses are often successful in spite of themselves and rather than because of themselves.
By not defining the outcome Tom wanted or expected, he was no different to the businesses Gavin talked about.
Tom wasn't making the most of the opportunities he had. He was wasting them.
A Business Hypothesis Approach
Before we could get to the process of creating and testing hypotheses, we needed to improve the presentation of data.
To do this, I set up some XmR charts to track the metrics Tom wanted to focus on.
We looked at higher-level metrics:
- Website Visitors
- Email subscribers
- Conversion rate
- Page views per visitor
- Dwell time
- Bounce rate
These charts revealed what I suspected. None of the improvements Tom had made caused the numbers to move outside of routine variation (If you don’t know about routine variation, you need to digest this: Becoming Data Driven, From First Principles).
We needed to move away from looking for potential explanations.
Conversely, our goal was to create a parameter meaning we had a provable explanation.
To do this, I told Tom we should change how he worked.
From now on, he had to write down every change before he made it. He would announce the anticipated results, providing us with enough evidence to determine the accuracy of the hypothesis.
Within each hypothesis, a clear value proposition would be evident giving Tom further justification to experiment. This framework would allow him to become a data-driven decision-maker.
It would bring an end to wasted opportunities.
The Building Blocks Behind The Theory
As I mentioned, for Tom to make this business hypothesis-led approach work, he needed a system to guide him through the different phases.
So I set this up as a simple database in Notion.
.webp)
Let me explain how it works:
The Null Hypothesis
You begin with a null hypothesis. It’s a statement of your baseline performance which we are saying can’t be changed.
The Alternative Hypothesis
The alternative hypothesis is a statement showing how we might disprove the null hypothesis in specific detail.
Over a pre-determined period, the conditions of the alternative hypothesis are put to the test. Providing there is sufficient evidence, then the null hypothesis is proven to be wrong. Importantly, if the alternative hypothesis isn’t proven then the null hypothesis stands until you create a second hypothesis to test.
“This is Epic!”
First, it was a beep.
A subtle notification telling me I had a message. Before I had a chance to read it, the phone started ringing. It was Tom.
“Oh my god!”
“It worked.” “What worked”, I replied.
“The hypothesis thingy you got me to write. It worked.” “This is epic!” Tom declared with delight.
Before he got too carried away, I took a breath and asked Tom to explain in more detail why he was so excited.
Tom explained breathlessly that his alternative hypothesis had been proven. Changing the primary heading and sub-title had increased conversion. Excitedly, he talked me through the null hypothesis.
It said: Readers subscribe to the newsletter via forms on the website at an average rate of 0.58%, with a routine variation between 0.4% and 1.2%.
The alternative hypothesis was as follows: If we change the text on the button from ‘subscribe’ to ‘try free’ then the conversion rate will increase to an average rate of 1%, thus changing the routine variation figures.
The results sustained over 4 weeks revealed an average rate of 1.2%.
Tom’s delight was clear. “Now I can tell what I changed and I have the proof to say with utter conviction, it worked!”
Concluding Thoughts
As Tom quickly discovered, there is no hiding with a hypothesis.
Thoughtless, unvalidated changes can create a lot of wasted opportunities. Before I introduced the Process Behaviour Chart, Tom wasted hours making changes he couldn’t substantiate.
Now, he could prove his theories with this experimentalist approach. A business hypothesis framework that quantified his ideas and objectively qualified his successes or failures. Subjectivity was from a distant past.
That’s why this approach works so well in a business.
Businesses are full of data. So entrepreneurs like Tom can become data-driven. They can make changes because they have hard facts to give them a precision level of appreciation for what is happening.
And if all this sounds a bit too structured and rigid, you might be right.
But does it work?
To find an answer I’ll direct you to a piece I wrote about the way Elon Musk makes decisions. In this article about feedback loops, you’ll find a hypothesis at the heart of the action. Here, the step towards progress is a test which creates a feedback loop.
Without the business hypothesis-led approach, Elon would not be where he is today.
An important hypothesis can develop longer-term thinking and create interest in planning a more thoughtful approach to the next steps.
That’s the real value for me.
Becoming data-driven — through a business hypothesis — leads you to meaningful change that will positively help your business.
Frequently Asked Questions
Cras tincidunt lobortis feugiat vivamus at morbi leo urna molestie atole elementum eu facilisis faucibus interdum.
What is a Hypothesis?
A hypothesis is a statement of your expectation of making a change. It's taking what you think will happen and framing it as an experiment to record the result.
How do you formulate a strong business hypothesis?
- Start by identifying a specific business problem or opportunity.
- Clearly define your independent and dependent variables. What factors are you testing?
- Craft your hypothesis as an if-then statement. For instance, “If we reduce response time in customer service, then customer satisfaction will improve.”
Why is hypothesis testing crucial for business decisions?
- Hypothesis testing allows businesses to validate assumptions before implementing strategies.
- By testing hypotheses, organizations can avoid costly mistakes and allocate resources effectively.
- For instance, before launching a new product, a company can test its assumptions about market demand through hypothesis testing.
What are null and alternative hypotheses in business research?
- The null hypothesis (H₀) represents the status quo or no effect. It assumes that any observed differences are due to random chance.
- The alternative hypothesis (Hₐ) proposes a specific effect or relationship. It challenges the null hypothesis.
- In business, these hypotheses guide statistical tests to determine whether evidence supports a change or effect.
What is a good example of a business hypothesis in action?
At the core of Elon Musk's decision-making process is a hypothesis. Every decision he makes, he seeks to test a theory which allows him to make progress. It is a fine example of a effective decision-making.
.webp)
