What They Are
Any decision boils down to choosing between different options. Sometimes the options are binary — either you’re going to do something or you aren’t. Other times they are discrete choices (do we go with A, B, or C?). Yet other times they are choosing from a continuous range (how much will we [charge, order, hire, etc.]?). In practice, even decisions on a continuous range typically end up being discrete choices (e.g. will we charge $1.99 or $2.99?).
A powerful step in these decision processes is to understand the conditions that would cause each of the options to be the right one. A decision tree is a flowchart that walks you through a series of conditions, creating a fork in the road at each condition. “If this is true, then we should choose between these options, but if it is not true, we should choose between these other options.” A decision tree represents the decision as the collection of conditions that lead to each of the possible decisions.
A very complementary tool is the simple question “What would have to be true?” (which I’ll abbreviate as WWHTBT for the rest of this article). You can ask WWHTBT about any of your options — specifically WWHTBT for this to be a good decision. I learned this question from Roger Martin. He calls this the most valuable question in strategy.
Specifically, Martin breaks this one simple question down into 7 questions that represent different beliefs about the industry, what customers value, competitive positioning, and competitor actions:
- What must we believe are the strategically distinct segments?
- What must we believe about how attractive the target segments are?
- What must we believe that the channel values?
- What must we believe that end-customers value?
- How must we believe our capabilities stack up with competitors?
- How must we believe our costs stack up with competitors?
- How must we believe our competitors react to our actions?
These questions can point us to good branching conditions in a decision tree.
When To Use It
WWHTBT and Decision Trees are most helpful when trying to align a management team to make a hard decision between relatively equally attractive options, especially in the face of uncertainty about the future.
But they can also be helpful earlier in the process. Asking the 7 questions within WWHTBT and creating the Decision Tree flowchart can help uncover additional options to consider. If you identify a branch point because it leads to a known option, the opposite side of the branch may represent one or more options you hadn’t even thought of.
Both WWHTBT and Decision Trees are also very helpful in the months and years following a decision. Tracking whether the things that had to be true are actually becoming true can help you identify a need to revisit your decision well before it becomes a disaster.
How to Use It
In his book Playing to Win with A.G. Lafley, Roger Martin lays out a 7-step process for “reverse engineering” a strategic decision by identifying “what would have to be true”:
- Frame the choice: identify at least two mutually independent options
- Generate strategic possibilities: broaden the list to be as inclusive as possible
- Specify conditions: For each possibility, specify which conditions would have to be true for it to be a strategically sound option
- Identify barriers to choice: Identify which conditions are most likely to not be true
- Design valid tests: Design valid tests to determine whether or not each option is worth considering
- Conduct the tests: Starting with the lowest-confidence conditions, test the hypotheses
- Choose: Compare test results and make informed choices
Of course, step 3 above requires thinking about the 7 component questions I identified above, and steps 4–6 involve a lot of work, so this isn’t as simple as it might first appear. Even if you don’t have the time or resources to rigorously follow this 7-step process, the framework can help identify and evaluate options for your decision.
Asking WWHTB and the seven component questions will likely also help you identify the different branches in your decision tree, including the different conditional branch points and the different resulting options.
A Decision Tree Example
As a simple example, I’ll go back to a fictional situation I’ve used in other articles. I helped launch Altimeter Software in 2016. We shut down the business in 2019, but if we hadn’t we would have faced some very challenging decisions when Covid hit in 2020. The company helped universities drive student engagement by encouraging participation in live events on and around campus. We had reasonable success selling to small Christian universities and had some interest from sororities and fraternities at large universities.
If the business were around in 2020, we would have had to determine what path forward would give us the best chance of survival. One of the decisions we would have needed to make is where to focus our software development efforts. We might have used a decision tree similar to that shown below.
Creating this decision tree started with the status quo — what would have to be true for sticking with the existing roadmap to be the right development plan? We would have to believe that small universities would still be able to buy our services and that the in-person events we enable would continue to matter.
Laying those two beliefs out as decision branches helps us identify other options.
If the small university market is no longer viable, what are our options? An obvious choice would be large universities. We had already had some discussions with fraternities and sororities at large universities, so one option would be to focus on that market. What beliefs would lead us to that being the best option? Focusing on fraternities and sororities would only make sense if we didn’t believe we could be successful selling to a large university as a whole. (In our business model, if a university bought our service, then all their organizations could use it, including fraternities and sororities.)
If the small university market is not viable and we believe we could sell directly to large universities, then we likely would need to make our product more acceptable to university decision makers, for example by integrating tightly into the leading University Management System (UMS) platforms.
If small universities are viable, but if live, in-person events are no longer a prominent feature of student life, then we would need to develop support for virtual events.
A really helpful exercise is to change the order in which you ask the branching questions. If I had structured this with the first branch instead being about live events, we might have developed this decision tree:
Changing the first question forces us to think more critically about our ability to support events that aren’t live in-person. What if the nature of our product becomes irrelevant when virtual events dominate? In that case, we would need to identify alternative markets to pursue. In 2020 George Floyd and Black Lives Matter became a dominant storyline, and there are other movements that people get passionate enough about to gather together in person and for which they want to get credit for being in “the right place at the right time”, so we could include that market as an option in our decision.
The real value of the decision tree is realized when the tool is introduced into a leadership team discussion. It provides a framework for having the hard, but very important debates about what we believe is happening and how we can participate in the future unfolding before us. Usually, it’s not too hard (but still important) to get people to agree on the options that make sense under different scenarios. The discussion often uncovers more options and more questions/branches — this is a good thing, within reason. The harder discussions are around which way the different conditions are likely to go.
Related Resources
Hopefully this article has helped you better understand Decision Trees and WWHTBT. The best resource on WWHTBT is Roger Martin’s book: