I’m in the middle of reading Daniel Kahnemann’s book, Thinking Fast and Slow. Half-way through, it seems as though intuition wins hands down. Many of us may like to think we make rational, logical decisions, but the evidence seems to rebut that assumption.
This casts a light on an experience from many years ago. In my consulting days, I used some pretty sophisticated decision-making tools and techniques to help clients evaluate their strategic options. But no matter how well I presented the data and how positive the feedback, I was constantly surprised by the number of clients choosing a path at odds with the recommendation. I concluded that there was no such thing as a good decision, only a political decision. The lesson I learned was to ensure the decision context incorporated the emotional and motivational elements of the relevant stakeholders and decision makers.
Notwithstanding this rather painful lesson, what also struck me was the different types of techniques that various leaders relied on. So, for all of you clinging to the hope that, in this post-fact world, there is still a place for rational decision making, here is a short list of the techniques I’ve seen used to make decisions. They are arranged in increasing order of sophistication.
The common characteristics for any decision are that:
- you must pick one option from several alternatives,
- there are some criteria you are using to assess their relative merits, and
- there are future events outside of your control (referred to as a “state of nature”) that will influence the assessment.
Throughout the blog I will use a simple example of a group of friends looking to choose a venue for a birthday party. The criteria are indoor capacity, outdoor capacity, location, décor, choice of drinks, choice of food, operating hours, choice of DJ, special features and price. The uncontrollable “state of nature” is the weather.
- Captain’s pick: in many ways this is the simplest technique. The birthday girl/boy makes the call. Others can make a recommendation, but my suggestion is, if the host is someone who likes to make a “captain’s call”, don’t get too invested in your recommendation. The main benefit is that decision-making is refreshingly quick in these scenarios. Note the birthday girl/boy may nominate someone else with perceived expertise e.g. someone that recently hosted a particularly successful party, then go with that call.
- Multi-voting: somewhat more democratic, but still relatively quick. Close friends may be invited to identify the pros and cons for each venue, put the options up on a board and then vote for their preferred option. This is not particularly sophisticated, but very straightforward and democratic.
- Multi-criteria: with this type of decision you assign a score to each alternative for each criterion. You can use “T-Shirt” size (S, M, L) or (High, Med, Low), Harvey Balls, numeric scores e.g. out of 10, whichever is easiest, but you must be able to aggregate the scores across criteria for each alternative. The alternative with the highest number is the winner. If you’ve ever sat through a session where this technique is the order of the day, you’ll recognise the issues it creates. It can get very tense as people argue endlessly about whether option A is 1% better than option B on criteria 2. For example, is the external dance floor for Venue A an odd shape, is it better to start at 8pm and finish at midnight or start at 9pm and finish at 1am. It also assumes that all criteria carry equal weight.
- Weighted criteria: this is probably the most common technique I’ve seen in practice. The panel assess the alternatives against the criteria, using a scoring system like the above but also assign weight to the criteria. For example, price might be more important than the availability of special features, so it’s given double the weighting. The scores for each alternative-criteria selection are then multiplied by the weighting (weightings across all criteria must add up to 100%). The aggregate score for each alternative is then derived. The issue with this approach is that the assessors are all weighted equally and there are some criteria that are subjective and hard to score e.g. décor if the choice is between red and blue.
- Financial decisions: if it’s just about the numbers, then payback, internal rate of return and net present value (NPV) are the most common techniques I’ve seen. While the purists may prefer NPV especially when trying to price an asset, for many of the investment proposals I’ve put forward the most common question from the Finance function is “What’s the payback?”, followed by “Is it self-funding i.e. can it pay for itself within the current financial year?”
These are the most common techniques I’ve seen in practice. They each have their advantages, and all have limitations, the major limitation being an inability to take uncertainty into consideration. Every NPV generated by a standard spreadsheet is fundamentally flawed. The output assumes that every variable and input is a fact. However, any forecast of future cash flows relies on a series of assumptions, which are, at best, a well thought through guess. Identifying a range of possible inputs would be far more suitable.
Another limitation is an inability to assess subjective criteria logically. What tends to happen in these circumstances is a tense debate with participants digging themselves into deep holes.
There are techniques that can overcome these issues. Some of them can get quite statistically intense, but in my experience, there was a far better understanding of the decision that was being made. This included an understanding of the importance of the risk factors, the attributes that could pivot the decision and the potential mitigants that could be employed.
So, the next time you need to make an important decision, why not try out one of these other techniques:
- Payoff matrix: this is a relatively simple tool. You calculate the payoff for each alternative for a range of possible conditions (this is bringing in the random element). In the party example, the venue with the largest indoor capacity may be more expensive and not as much fun, but better than being outside on a cold-wet night. Based on how optimistic you are about the weather you can then choose your course of action.
- Expected monetary value: an add-on to the payoff matrix described above. You could assign probabilities to the likelihood of good or bad weather conditions and then calculate the expected monetary value. Assigning the probabilities may be tricky, but you could pay a meteorologist/expert to help.
- Decision trees: in most cases there are several decisions to be made. In the party example, if the weather forecast closer to the date of the party is looking bad, you may make different decisions e.g. hire more patio heaters. Using a decision tree to represent this type of problem is a good way to visualise the possible outcomes. The payoff is given at the end of each branch.
- Utility models: of course, with a party it’s not just about the cost. The magic ingredient is the guest list and the ambience. Menu and drinks choices, location, date, start time etc. all play a part in determining who will accept the invitation. This is where utility comes in. You may be able to get a cheaper option, but if that dissuades your guests from accepting then the financial saving detracts from the overall value. Using utility models is particularly helpful at factoring these elements in – especially risk.
- Monte Carlo simulation: a great way to enrich a financial model. For the key variables, rather than entering a single number into the spreadsheet, you enter the probability distribution characteristics. For example, if the useful life of an asset has a mean of 5 years, a standard deviation of 2 years and follows a lognormal distribution, you enter these parameters (there’s specialist Excel Add-Ins like @Risk to help). The spreadsheet will re-calculate thousands of times, each time selecting a number to place in the input cell from the distribution based on its likelihood. Every time the input is changed, the output is calculated e.g. the NPV. At the end of the simulation, the spreadsheet will produce the range of possible answers for the NPV and their likelihood – this is a far better and more realistic answer than a point solution.
- Analytical hierarchy process (AHP): finally, the AHP is an approach for choosing between alternatives where the criteria are quite subjective e.g. is choice of colour more important than overall design. The AHP approach allows each stakeholder to state their preferences by initially comparing each pair of criteria and assessing how much better one is than the other. This pair-wise approach allows you to be consistent. For example, if I believe location is far more important than the drinks menu and the drinks menu is more important than the décor, then the location must be more important than the décor. After assessing the criteria, it’s then time to rate each alternative against each of the criteria using the same principle – pair-wise comparison. Once again, specialist software helps e.g. Expert Choice. This is a far more robust and collaborative approach to decision making than any other technique I’ve seen.
Hopefully, you’ll have a better understanding of the choices at hand. Of course with the momentum behind big data and analytics, there are now no excuses for not trying these techniques. And remember, any decision involves risk, there is no such thing as perfect information and “do nothing” is always an alternative.
For those of you wanting to find out more I’d recommend the following three books:
Making Hard Decisions: Robert T. Clemen
Spreadsheet Modelling and Decision Analysis: Cliff Ragsdale
Financial Models Using Simulation and Optimization II: Wayne L. Winston