Decision Trees
Decisions are rarely as linear as making a big yes/no decision and then following the action before discovering, at the end, if you were right. Decision-support techniques however are often presented in this simplistic way. The emphasis is on the initial choice of action with little said about what happens next. This is fine in business school, but in the boardroom a tool for handling the messier process of real-world decision-making is more useful. An underused but surprisingly effective technique for this is a decision tree.
Decisions
The truth about real-world decisions is that there is rarely a clear picture of what the alternatives are. Future problems are dimly perceived, possible solutions hazily understood, and the likely outcome of actions either unknown or unknowable. Moreover, if you wait until all becomes clear you will still be waiting when the liquidator walks through the door. In this environment there are times when you place big bets, but often there will be much subjective judgement, small bets and simple trial and error that comes before.
Decision Trees
How do decisions get made? It’s a mix of informal discussion, formal meetings, investigations and recommendations. Much of the risk and uncertainty is often held in people’s heads or described in written text. A decision-tree is simply a way of codifying this in a logical and visual way. As long as you avoid it looking like something a computer programmer would produce, it’s a way of simplifying confusing situations.
I’ve often found clients appreciate decision trees more than I expected. It has seemed to be a helpful way of expressing a decision they had reached intuitively but the decision-tree gave them language to express the reasons why. Here are some instances in which I’ve used them.
Early-Stage Investment / R&D Investment
The key to high-risk, early stage investing is not to put all your chips on the table too soon. Before investors go ‘all in’ they prefer to dole out the money in bite-sized chunks and on condition of reaching specific milestones. It helps to limit risk and keeps the company’s management focused on the investors’ priorities. Too much money too early carries a big risk that it will be squandered. A common example of this is spending time building a website and designing brochures instead of going out and getting the first customer order.
The examples below are taken from a portfolio of investment opportunities pitched to a Dragon’s Den-style panel of venture capitalists. Exactly the same technique could be applied to R&D decisions which are, in effect, an extreme form of early-stage investing.
Contingent Situations
Decision-trees are perfect for multi-layered situations where the direction of travel is contingent on other, sometimes unconnected, decisions. These need not be framed in financial terms. Most of the decision points in the example below are non-financial judgements and there is a considerable difference in the possible outcomes, depending on how the questions are answered.
Follow the Breadcrumbs
Sometimes there is no real business decision to be made in terms of judgement to be exercised but it can nevertheless be helpful to illustrate a process using a decision-tree. In such circumstances the ‘decisions’ are more ways of interpreting rules, but the rules are complex and so it’s helpful to be guided through them. The example below was used to interpret a complex VAT situation in a not for profit organisation.
Words of Warning
Decision trees by their nature are a presentation technique. Unlike other decision-support techniques they won’t tell you what to decide in any given situation.
Summary
Decision trees are a way of describing a complex and contingent environment in ways that help the decision-makers arrive at a course of action.