Reimagining Forecasting in Uncertain Times

One of the key purposes of forecasting is to help organizations continuously anticipate the future, assess its likelihood, consider the implications and use all the available information and practical techniques to make confident decisions that maximize the potential of the business and improve enterprise performance.

Of course, this is not an attempt to simply predict and control the future as no one is capable of predicting the future with certainty. Instead, it is about building an effective and efficient process that enables decision makers to evaluate alternative courses of action available to them to respond quickly to known knowns, known unknowns or unknown unknowns when they happen.

The novel coronavirus pandemic which started in Wuhan, China and rapidly spread across the globe has killed thousands of people, caused world’s markets to plummet, disrupted global supply chains, and forced non-essential businesses to temporarily or permanently shut down.

Operational and financial forecasts that a few months ago painted a rosy picture of the future have had to be thrown out of the window because of the Covid-19 crisis. Companies have come out in droves and announced revised earnings and profit estimates.

Several governments, academics, research bodies, businesses, and other organizations were caught unaware by this outbreak – demonstrating our inability to predict the future. Considering our inadequate forecasting capabilities, should we therefore abandon forecasting completely? Can we confidently say forecasting is a waste of time and resources? The answer to both questions is a resounding ‘No’.

In order to make confident, reliable and timely decisions, decision makers require information about the past and information about the future. Absent this information, or it is deficient or misleading, then decision-making on key organizational performance matters is no more than guesswork.

Unfortunately, in today’s increasingly unpredictable, uncertain and rapidly evolving world where things can happen at lightning speed, anchoring important future business decisions on historic information alone can misinform decision-making resulting in lost opportunities and catastrophic business failure.

While it’s imperative to understand business trends, it’s also important for management and decision makers to appreciate that they cannot simply rely on the past to guide them in the future given their role is to make the future different to what it otherwise might be.

Forecasting is not supplementary to the annual budget

The traditional annual planning and budgeting process is certainly fraught with many shortcomings and for most organizations it is completely dead. Outdated and meaningless. Most businesses spend several months each year agreeing the budget and then monitoring actual performance against it.

They plan for a desired future in order to make it come about. Any deviation from budget or the initial plan is regarded as a variance and is therefore wrong. But many times, changes in the business environment invalidate the assumptions on which they based their original plan and quickly render the budget obsolete.

The budget is next updated with a forecast with a horizon that declines as the reporting calendar moves towards the fiscal year end. As a result, decision makers have limited visibility about the company’s future considering they are ill-equipped to look beyond the twelve-month planning horizon.

In one of his CIMA FM Magazine articles If the Coronavirus Outbreak Disrupts Your Budget, Bjarte Bogsnes reminds us:

In the world of business today, there is more on a long list of things that we can’t control. The only thing we know about our budget assumptions for next year is that most of them will be wrong.

In such an unpredictable world, we shouldn’t expect budgeting to produce predictable results. It assumes that we can sit down in the autumn and decide everything for next year — what to earn and what to spend and invest, all laid out at the lowest detail level. We believe this gives us control. It gives us nothing but an illusion of control.

CIMA FM Magazine, March 2020

So unlike in budgeting where we consider a single desired future outcome, in forecasting we know the future might not come about because the assumptions might be wrong or have been changed. Thus, instead of continuing on the same path based on existing assumptions it’s imperative that we quickly change course if we are to avoid a catastrophic iceberg crash.

Remember, the purpose of forecasting is not to predict the future with certainty but rather support decision-making. A good forecast highlights a projection of the future with some ranges around it, an informed and reliable explanation of what is driving uncertainty and a cogent plan for the business to mitigate emerging risks or exploit the opportunities.

Talk about possibilities and probabilities, not certainties

In Super Forecasting: The Art and Science of Prediction, Philip E. Tetlock and Dan Gardner shed light on the dangers of organizing our thinking around Big Ideas, whether true or false, when making forecasts. In this scenario, we tend to force complex business problems into preferred cause-and-effect templates and treat those that fail to fit as irrelevant distractions.

As a result, we get awfully confident and less reluctant to change course even if changes in the external environment are clearly nudging us to do so and have invalidated our initial assumptions and projections.

Since we never pause to evaluate whether the evidence at hand is flawed or inadequate, or if there is better evidence elsewhere we suffer from an illusion of knowledge or what Daniel Kahneman termed WYSIATI – What You See Is All There Is.

The problem with zoning on one Big Idea is that it distorts and doesn’t improve our foresight. Instead of viewing and processing new information with a fresh pair of lens, the additional information received is treated as unhelpful because it’s all seen through the same old pair of tinted glasses.

In forecasting, there are no certainties but rather possibilities and probabilities. To be able to produce accurate enough forecasts, it’s critical that we deploy not one analytical idea but many and seek out information not from one source but many. Then consider and aggregate alternative views.

Generating different perspectives (that is coming up with an outside view and inside view) of the business and integrating the two isn’t the end but a good beginning. It helps us understand what other forecasters think, and also what outside and inside views they have come up with.

Nonetheless, since teams constitute of individual members with different educations, training, experiences, and personalities – a smart leader should not expect consensus of opinion at all times but must treat its appearance as a warning flag that group-think has taken hold.

On the contrary, a display of differing judgments should be welcomed as evidence that the people around the table are actually thinking for themselves and offering their unique perspectives.

Forecast, measure, revise. Repeat

How predictable something is depends on what we are trying to predict, how far into the future, and under what circumstances. The further we try to look into the future, the harder it is to see.

In business, finance organizations rely on robust models to churn out short, medium and long-term financial estimates on company performance. And most of the time after forecasts are produced we hardly measure our forecasting performance and improve the quality of our forecasting process accordingly.

All forecasts contain some level of variation or unsystematic error, but a reliable process maintains this variation at acceptable levels for the purposes of the decision business leaders need to make. Without measurement, there is no revision. And without revision, there can be no improvement.

Certainly, this does not imply setting arbitrary targets such as plus or minus 5%. This approach of measuring the difference between actual and forecast outcomes and expressing that as a percentage is flawed. One of the reasons being that errors are assumed to be evidence of poor forecasting, and success in meeting the numbers is viewed as good forecasting.

Forecasts must have clearly defined terms and timelines. People attach very different meanings to vague verbiage like “significant market share,” “certain,” “seriously possible,” “a fair chance,” and “likely.” Such ambiguous language renders forecasts untestable. The same holds true for forecasts (economic) that claim undoubtedly that something will or won’t transpire in future but fail to explicitly define the time frame.

In their work Future Ready: How To Master Business Forecasting, Steve Morlidge and Steve Player further highlight the importance of comparing ‘like with like’ when measuring forecasting performance. That is, we should not attempt to compare the results from forecasts produced using different time buckets.

For example, comparing the actual for Quarter 1 with forecasts made at the end of December, January and February. Neither can we compare a forecast for Quarter 4 made in January with a forecast for Quarter 4 made in June. Although the time buckets are consistent, the forecast lead times are not the same.

We therefore should measure forecast error within forecast lead times using consistent buckets and consistent forecast lead times.