In recent years, business forecasts have had to become faster and more flexible-ready to be adjusted at a moment's notice. As earnings and product cycles accelerate, it has become clear that the business world is decisively moving towards treating rolling forecasts as standard. A slow process of budgeting and forecasting is no longer enough.
This digibook contains a step-by-step guide on how to meet stakeholders' expectations, with case studies and examples of how smart CFOs are adapting their processes to improve the quality and timeliness of their business forecasting.
In this guide, you'll find out:
If we seem to be missing three big ticket items, it's because they're covered elsewhere in this series. Check out our three digibooks on Thriving in the Digital Age, Organizational Change and Reporting
Who will find this digibook useful?
This has been written with CFOs and finance VPs at international companies of all sizes in mind, but our research and case studies also include smaller businesses and public sector bodies, as well as larger multinationals, so any senior finance person will find it useful. It also has insights into the life of the CFO that other C-suite team members may find useful in understanding the challenges and changes taking place in forecasting and re-forecasting.
Introduction: The Business Forecasting Landscape Post-Crisis
Over 62% of organizations find that their budgets reflect a single point in time and quickly become irrelevant.KPMG and ACCA survey1
The traditional process of annual budgeting and forecasting-compiled in the three months before its release, and based on static assumptions-started to look pretty silly in 2008. When the global financial system collapsed, business plans and cash flow forecasts ended up worthless, and many companies had no process for re-forecasting on the run.
Forecasting has changed dramatically since then, driven both by further financial shocks in Europe and China, and by the rapid rise of better technology and richer data.
CFOs are expected to be able to reforecast in a window of just a few days, all the while sifting through unprecedented levels of information. It's a big challenge-but it also comes with big rewards.
Companies good at forecasting saw overall share prices rise a third more than other companies.EIU/KPMG: Forecasting with confidence2
Most CFOs are moving towards something very close to a true rolling forecast; that means they are able to create forecasts that are:
The average forecast cycle has fallen to just two weeks long; the smartest companies have driven this down to just three days.3
Budgets are now understood to be flexible: torn up every few months, in response to an always-on forecasting process that constantly updates plans.
Forecasts are no longer based on past results: category growth, market share, human capital, customer satisfaction, and a range of other metrics are fed into the system, making it possible for a prediction to respond instantly to fluctuations in the marketplace or the workplace.
That said, speed and new data are not magic wands that you can wave and instantly have all your problems solved. An astonishing 80% of companies believe their forecasts are unreliable.4 To put yourself in the 20% that are getting it right is the biggest challenge of all.
In this digibook, we'll look at how smart CFOs are building rolling forecasts and mitigating the dangers presented by complex data and quick processes. At the end, there's an opportunity to build a personal checklist for meeting these new challenges head-on.
Three things to throw away when building your rolling forecast
The budget process 'sucks the energy, time and big dreams out of an organization' according to former head of GE, Jack Welch, who claims that companies succeed not because of it but in spite of it. Ever more CFOs would seem to agree with him: it's no longer about 'dropping off a cliff' at the end of the year.
It's been five years since Unilever abolished its annual budget in 2010 and adopted an eight-quarter rolling forecast. This has provided the visibility and flexibility to adapt rapidly to new competition and has removed the need to spend six months putting together an annual plan that was out-of-date by publication.7
As Unilever found, adopting a rolling cycle turns your forecasting into a dynamic review and planning process that runs at the ever-increasing pace of the business.
So, what are the key points to avoid when implementing a rolling forecast?
The annual plan
Annual plans lure you into a false sense of precision simply by virtue of the time it takes to prepare and the level of detail included. A plan represents a single-point estimate of future outcomes, and is therefore a terrible way of predicting changes in a complex business. By abolishing the plan, you can reallocate your resources to what really adds value.
Prediction as an end rather than a means
Budgeting is not the goal. It's a means to an end. The purpose of forecasting is to influence the future, not predict it. Your budgeting needs to provide not one definitive answer, but rather a set of actionable recommendations based on the most probable set of scenarios.
Looking backwards rather than ahead
Basing your analysis on trends and averages from last year is the mistake that made CFOs look most foolish in 2008. You need to model future scenarios based on current drivers, so that the impact of changes in the marketplace can be clearly expressed, and the range of outcomes over the next 18-24 months predicted.
Euroclear chooses to forecast across 18 months, refreshing the forecast at least three times a year, allowing it to anticipate increase in new business and understand how this will affect cost.8
Tapping the wisdom of crowds
The overused phrase 'the wisdom of crowds' originally referred to a very specific statistical quirk: in the right circumstances, large groups of people are excellent at predicting future events-to the extent that an average of ill-informed guesses is frequently more accurate than an expert calculation.
The statistical analyses behind this process are fascinating (though beyond the scope of this digibook), but they boil down to understanding which sets of opinions will lead to a bell curve.
As a CFO in the digital age, you have for the first time direct, unmediated access to frontline employees in your organization. And that means that you can collect their opinions to make predictions that combine remarkable accuracy with great speed.
The process starts with the competitive testing of forecasts: run (for example) a monthly sales forecast the traditional way, inputting all the drivers, checking in with all your VPs, and deriving the data via some careful algorithms. Then run it again, by asking your frontline staff to estimate how much they expect to sell next month. If the latter process performs better than the former (and you'll need to do this a few times to confirm), you may have found a way of speeding up, simplifying and improving accuracy.
The system will only work if you can separate the forecasting and incentive processes. One of the biggest challenges of getting intelligent data from a wide amount of people is avoiding 'budgeting politics'. If targets, measures and rewards are linked to the numbers in a forecast, it becomes an exercise in self-preservation and self-aggrandizement.
Borealis, the Danish petrochemicals company, separated targets and rewards when it implemented a rolling forecast, moving them into the hands of a quarterly review committee. It very soon saw an adjustment of forecasts towards a more realistic view of essential project expenditure.9
Finally, you need to design a lean process for people to give you their data. If you are asking 100 people in your business for a forecast once a month, they need to be spending five minutes on filling out an online form, not five hours fiddling with spreadsheets.
Four steps to ensure Business forecasting Is relevant
Just getting yourself into the habit of rolling forecasts isn't enough. Is it relevant? Does it tell you what you need to know, or what you've always assumed you need to know?
The challenge of dropping big data into a driver-based system is knowing what you should be reporting on. Too many businesses are wasting their time producing 'write-only' documents and metrics: non-value-add filler, which nobody in a line of business ever reads or acts on.
A.P. Moller-Maersk's rolling forecast contains only 35 input accounts and three to seven KPIs; its previous annual budgeting process was built on inputs from over 1,000 of the group's legal entities.10
Streamline: complexity is the enemy within
A simple response is always the best answer to a complex problem
This goes for your forecasting processes as much as for anything else.
Simplifying how your teams approach the forecasting procedure is a key method of driving efficiency- and efficiency is a key driver of value.
The first step to streamlining is to standardize . Your inputs need to follow common standards and classifications. This is particularly important in allowing you to re-forecast, as changing one part of a modelled scenario does not mean changing everything else if the input follows the same commonalities. Hyatt's 'Project Unify' standardized into one dashboard to take them out of 'spreadsheet hell' where no-one but the creator understood what was going on, into a position where progress could be easily tracked across all areas.
You then need to build on this by centralizing . To cope with the challenges of scaling up the number of inputs from across the organization, forecast and planning processes need to be put into one (virtual) place to ensure that everyone is working to the same standard. Replacing complex spreadsheets with collaborative technology, for example, facilitates easy and quick organization-wide input. Not only can this sharply reduce response times, it reduces the likelihood that input errors will go unnoticed.
You can take this process further by outsourcing or automating routine functions. The finance team's ability to reforecast depends on agility, and this disappears when they are bogged down in managing routine processes. The best way of identifying these is by asking yourself and your team what the dullest part of the job is: if it's really boring and needs to be done more than once a quarter, it can probably be managed by a clever SaaS application or a sophisticated back-office support service.
Finally, track your failures . There is always a good reason why a prediction went wrong "this time". There were currency fluctuations, or a change of regulations, a huge deal went through, or the new product had to be recalled. So it's worth looking back over the forecasts you have made, and checking which ones actually go wrong every time. These are not just a waste of your team's time-they are actively harmful to the business. Sometimes they can be fixed by changing the inputs or switched to range forecasts; but sometimes, you'll need to admit that forecasting is always hard... and there are bits of the business for which it is actually impossible.
Use this checklist to build your action plan. As you select each item, they will build into a comprehensive set of next steps for you.
Which of the following do you need to do?
Check those that apply
Your action points
Building a rolling forecast
- Throw out your annual plan
- Treat prediction as a means, not an end
- Switch to driver-based forecasting
Draw on the wisdom of crowds
- Run competitive tests to see if crowd-sourcing provides better data than your current system
- Separate the forecasting and incentive process
- Design a lean process for getting data from front-line staff
Make your forecasts relevant
- Identify conversion factors
- Eliminate non-value-add detail
- Consolidate your metrics
- Use range forecasting
Streamline your processes
- Standardize metrics and classifications
- Centralize your data
- Track your failures over time