Working In Uncertainty
Illustrative improvements to ways of working
One important way to improve risk/uncertainty management is to make changes to the way you work. Applied to an organization this should lead to to incremental changes to core management activities and basic business processes so that they work better in the face of limited knowledge and control. To a larger extent, risk/uncertainty are managed within core management activities. This idea is not really new and it's not rocket science.
I've found that people are quite good at thinking of ideas for improved handling of uncertainty, but of course with more knowledge and practice they get better. Exactly what incremental changes you make is up to you, but here are some suggestions to show the sort of thing that will usually be worthwhile. You probably know some or even most of them already, but perhaps will find some new ideas you could look into further another time. All these ideas are well established or too obvious to need publishing. Some may not be normal practice for most organizations, but there are plenty of organizations doing them already with success.
By considering these examples you should be able to get a sense of how big the opportunities for improvement are in your organization, and organizations generally. Once you understand that we tend to underestimate the limitations of our knowledge it is obvious that there are opportunities for improvement all over the place, just waiting for someone to notice their value.
By 'core management' I mean things like making strategies, monitoring progress, delegating, planning, and so on. It does not include 'risk management' or 'internal control'.
By 'ways of working' I mean processes, techniques, tools, systems, skills, and anything else you can think of that could be changed to improve performance.
Here are some typical examples of core management activities and potential improvements.
By forecasting I mean making any prediction about what might happen in future if we try to take a course of action, or perhaps regardless of our course of action.
Predictions occur in a wide variety of management activities. For example:
Accounting forecasts, perhaps estimating what financial results will be at the financial year end.
Market forecasts, perhaps used to inform strategy making.
As a thinking step in planning. E.g. What would happen if we did this? What about if we did that instead?
As a thinking step in design. E.g. What would happen if we made it like this? How would it perform? What would happen to sales? How much would it cost to make?
Our usual bias, as human beings, is towards thinking we are better at forecasting than we really are. We also tend to think we have more control than we actually have. This bias often gets built into management practices. Consequently, we make poor decisions and tend to be caught out by the unexpected more often than we should be.
Specifically, we often use forecasts that are just the self-serving guesstimates of several people added up, we rely on forecasts that give a best guess for important numbers without giving any information about how certain the forecast is, those forecasts are based on best guesses for inputs (which means they are are affected by an important mathematical error known as the Flaw of Averages), and we fail to consider the value of being able to respond to what is learned in future.
These faults, and others, mean that we under-estimate the value of flexibility, precautions, and other responses to uncertainty and work with best guesses that are systematically wrong.
There are various ways to lessen these problems. These include:
Use scenario planning, which involves thinking about how unexpected situations might arise.
When getting expert judgemental input for predictions, use debiasing protocols devised by psychologists.
Add prediction intervals to point forecasts.
Analyse chains of cause and effect and present forecasts in terms of multiple possible outcomes.
Use spreadsheet designs that eliminate the Flaw of Averages and enable the value of keeping options open to be assessed.
Create spreadsheet models where alternative courses of action can quickly be tried out during planning.
Drive forecasts with real data instead of just gut feelings and wishful thinking.
Use simple simulations to understand how events could unfold even though specific predictions are impossible.
Learning from experience also appears often in management. For example it occurs in:
project management, when something is delivered to stakeholders that they can use and give feedback on; and
manufacturing and service delivery.
As over-confident forecasters we tend to think there is less scope for learning than actually is the case. We often don't plan to gather information or look at it, we assume our initial plans will work as predicted, and we are slow to realise that ideas are more or less effective than we expected. Often, people set priorities but do not plan to revise those priorities, which is usually a sign of overconfident prediction.
We tend to think of learning from experience as either something that happens without effort, or as something done by analysing numbers and reaching conclusions. Neither is realistic.
Things we should often do more include the following:
Plan to gather information, learn, and revise priorities; think of plans as series of trials.
Gather richer data more quickly, including information about the immediate effects of actions, not just their eventual impact on money.
Observe causality directly by looking at individual cases in detail, rather than trying to establish causality from correlations alone.
Use controlled comparisons more often, including multivariable experiments.
Account for factors we understand to discover more about the influence of factors we don't understand.
Deliver projects incrementally, because delivering something useful to stakeholders usually produces much more learning than another stage of paperwork.
Test usability – of everything from advertisements to speeches to software...
The common factor here is creating some words that are designed to be followed in future, when possibly circumstances have changed in ways we have limited knowledge of.
This occurs in, for example, strategy making, delegation, and contracting (both in sales and purchasing).
The effect of putting plans into writing, giving instructions, and making contracts, is that we tend to reduce our freedom to adapt in future. We take the sort of simple, one-path-to-one-destination style that works for everyday requests like 'please pass the salad' and try to make it work in situations where there is considerable uncertainty. Metaphorically, we may not know if there is salad on the table, if the other person has a hand free to pass anything, or even if there will be a table when our instructions come to be carried out.
Changes to methods that can help us deal more cleverly with these limitations include the following:
Include more contingencies (i.e. "If...then..."), which is a natural consequence of scenario planning, for example.
Create plans with policies, which are in effect standing contingencies.
Give flexible instructions using contingencies and value models (i.e. express what you would value and leave others to give the best value they can).
Write flexible contracts, for example providing for controlled renegotiation of details.
At its most general, establishing objectives/values is about trying to work out what is in our interests. What would make us happy?
It occurs on a large, slow scale in strategy making and annual planning, but also from moment to moment and when we are planning our day. It's part of formulating design briefs, setting up projects, and making contracts.
Some think this is something that comes from within. You just have to decide what you want and then work out how to get it.
In reality, facts matter. We are not good at knowing what will really make us happy. As individuals we are not good at predicting how happy things will make us, we misjudge the appeal of holidays, gadgets, friends, clothes, and so on. As a species we clearly have difficulty knowing what will make us happy. Many thought it was money, luxurious consumption, and sports utility vehicles. Unfortunately, that now looks like something we cannot sustain and that will lead to great hardship for future generations. In fact research suggests that volunteering to do some good for someone else makes most of us happier than luxury.
The truth is that 'goals', 'objectives', 'targets' and the like are just temporary planning assumptions that we should be ready to revise when we discover something better.
Practical consequences of being too slow to improve our understanding of what is in our interests include:
Continuing to work towards obsolete targets.
People gaming with targets because these are unchanging and fail to capture everything that is important.
People getting insufficient guidance from objectives because they express how only one particular outcome would be valued.
Techniques that can improve performance in uncertainty include the following:
Identify uncertainties about how actions lead to outcomes and so on to results of real value. Explore and research them over time.
Move from expressing target levels to giving values for a range of possible outcome levels.
Develop value models, which take this further with models (often quantified) that express how much you value different levels of performance on a variety of performance characteristics. Value models are far more informative than targets.
Revise objectives more often, moving from annual to quarterly, monthly, or even weekly. Many small adjustments made quickly and routinely replace lengthy processes that lead to larger but infrequent upheavals.
By 'basic business processes' I mean the main work of an organization — whatever its sector — such as buying, selling, making things, book-keeping, serving customers, and so on. This is the work that the management activity is supposed to be managing. There's a grey area between these two but you just have to choose a place to draw the line.
Ways to improve these business processes so that they perform better under conditions of uncertainty are well known to many people. These usually focus on reliability and fraud problems. Think of auditors checking controls over financial reporting, or production engineers trying to find a better maintenance schedule to keep machines running. The uncertainties are usually around when and how components of a system will fail or be attacked.
Hundreds of people receive notification of new publications every month. They include company directors, heads of finance, of internal audit, of risk management, and of internal control, professors, and other influential authors and researchers.
Please share: Tweet
Words © 2011 Matthew Leitch