Hypothesis trees: the basics

A hypothesis tree is a powerful tool for top-down problem-solving in business. 

It's used to help teams target a problem and break it down into sub-hypotheses to make it easier to solve. Think of hypothesis trees as a shortcut—teams create a theory about what's causing the problem based on their experience and use it to form a hypothesis on how to go forward. 

Hypothesis trees are famously taught in McKinsey Mind as a way to solve a problem in a single meeting. For McKinsey analysts, hypothesis problem-solving usually follows three simple steps: 

1. A hypothesis is formed about the problem and how to solve it
2. Data is gathered to prove (or disprove) the hypothesis by comparing results
3. The original hypothesis is updated when the data has been analyzed, and the opinion is proven or discarded

Teams may also use hypothesis trees to get an even better idea if a hypothesis solves a problem. These trees have three main characteristics: 

1. A primary hypothesis to answer the question 
2. Structured layers (sub-hypotheses) to add nuance to the hypothesis (e.g. the first layer contains supporting data for the hypothesis)
3. Each layer and issue uses M.E.C.E to examine and support the previous hypothesis

For teams using hypothesis trees, there are two huge benefits. 

First, they reduce the time taken to solve a question as the discussion starts with an answer that gets proven or disproven. Second, teams are forced to give an educated guess upfront to the hypothesis, which captures their experience and insight to speed up the problem-solving process.  

How a flawed hypothesis brought down Blockbuster

Let's bring a hypothesis tree to life by examining the core question Blockbuster CEO James Keyes faced when he tried to resurrect the business in the 2000s. 

Keyes asked himself: How can Blockbuster increase the profitability of its stores?

The primary hypothesis he worked with was Blockbuster's existing strategy was working, and doubling down on it would increase store profitability. This hypothesis was reached as a result of team experience and their understanding of the core problem. 

With the primary hypothesis defined, Blockbuster created three supporting hypotheses to save money and increase profitability: 

• Reduce investments in Blockbuster online
• Maximize cash flow by increasing late fees 
• Add impulse purchase products and stores

It is easy to see how Keyes assumed data and facts to reinforce his assertions with these three additional hypotheses in place. 

For example, the hypothesis behind adding impulse purchase products and stores was based on the data that customers spend an average of 20 minutes in the store when picking out a film. Blockbuster assumed customers would be tempted to buy candy and popcorn within this timeframe. 

The danger of this example is the initial hypothesis—it's flawed. Blockbuster's existing strategy wasn't aligned with shifting customer expectations (going from physical DVDs to streaming), not to mention the fatal blow of cutting investment in Blockbuster online. 

The incorrect hypothesis devastated Blockbuster and the company went bankrupt in 2010.

The three common pitfalls with using hypotheses to solve business problems

Using hypothesis trees for problem-solving can save time and draw out your team's expertise—but they also come with great responsibility. 

A common problem with hypothesis trees is creep. As teams lean on their expertise and experience to create hypotheses, they are exposed to narrow core question definitions and limited alignment. 

Then comes the biggest risk to hypothesis trees—confirmation bias. 

Let's use an example—congestion in the streets of Los Angeles—to highlight the pitfalls with hypothesis trees in problem-solving.  

The problem's primary hypothesis could focus on reducing the costs of tunnel boring and moving traffic underground. However, that hypothesis also represents a very narrow view. So, another hypothesis could be investing in self-driving vehicles above-ground and public transportation, or encouraging businesses to let their employees work from home to reduce traffic. 

Teams must also consider limiting team alignment. Since the hypothesis is forced early on, it doesn't let the data speak. One hypothesis must be picked, and it limits opinions or data on other hypotheses being discussed and can impact the team's ability to solve the problem successfully. 

Finally, hypothesis trees are prone to confirmation bias. Humans have over 180 cognitive biases that our brains use to guide decision-making. But psychology experts say these biases are bad for business, as they can lead to incorrect decision-making, and teams can default to embracing group think. Hypothesis trees start with the strongest solution in our brain, and we are wired to believe it's correct because of our experience and expertise—which can be a disaster for business. 

Expert Corner: Rob Jenks' pros and cons of using hypotheses to solve problems

Rob Jenks is the senior vice president of Corporate Strategy at Tanium, a cybersecurity and systems management company.

His role has no room for bias, and every decision he makes has to be rooted in data and evidence. 

According to Jenks, to overcome a structural problem it's best to break it down into parts and find bite-sized chunks that can be solved. When you piece it back together, the chunks will either prove (or disprove) the hypothesis you've been working on. 

Jenks says it's important to remember a hypothesis is a provisional answer and a theory that needs to be investigated. 

"The reason it works so well is that it makes the solution transparent and when you identify that last leaf on the hypothesis tree, you realize the subpart of the problem often is trivial," he says. 

"I remember from my McKinsey days. We would always ask teams literally on the first day— what's the answer to this problem? What's your day one answer? What you're really saying is: what's your hypothesis?"

Even when a hypothesis is identified, Jenks says the pros and cons of using hypothesis trees to solve problems can't be forgotten. 

The bad

The downside to using hypothesis trees, Jenks says, is you simply could be wrong. 

"We might end up going down a path that's driven by the hypothesis that ends up being completely wrong," he says. "You can waste a lot of time as hypothesis-driven problem-solving is sometimes just confirmation bias."

The good

Hypothesis trees can be a very effective and rapid method of solving problems. 

"When you have an answer in mind, It's much easier (and more concrete) to then think about it and how to prove the hypothesis," Jenks says. "It dramatically accelerates the top-down structuring of problems, and it focuses on the acceleration to the answer and it cuts off extraneous work."

Use hypothesis trees wisely to speed up your problem-solving

Hypothesis trees are a powerful tool to speed up problem-solving—if they're used correctly. 

Hypotheses are born out of expertise and experience. Backed up with evidence and data, they give businesses a head start when solving problems and can significantly speed up the process. There's a reason firms like McKinsey use them in daily interactions with their clients—they work. 

However, there's a downfall to relying on hypotheses to solve problems—the human brain. Our brains are wired in cognitive bias. If we think we know the answer to a question or problem, we can forget that we need to back it up with facts. Missing this step when using hypothesis trees can lead to incorrect problem-solving and wasted time. 

Our advice is to use hypothesis trees as they were designed: as a provisional answer to give your team a head-start to solve complex problems.

Now you know all about hypotheses trees—but have you heard about issue trees? Learn more.