There’s one sure way to kill rolling forecasts: Have people think it’s like budgeting but must happen more quickly and more frequently. Nothing is further from the truth. This lesson will cover best practices for implementing rolling forecasts.
Companies with Better Odds
Certain types of companies are better positioned for a successful implementation. They are:
- Companies in volatile and uncertain markets
- Companies comfortable with constant change and growth
- Companies frustrated with outdated planning information
- Companies that have investor pressure for frequent accurate forecasts
- Companies with a variety of management processes for resource allocation, performance management, and KPI tracking
These types of companies are most hindered by traditional budgeting. They also have other processes in place to supplement rolling forecasts, especially if traditional budgeting is discontinued. Some of these items may need to be put into place or strengthened before implementing rolling forecasts. Company cultures resistant to change may require slower implementations.
Building Support for Implementing Rolling Forecasts
The first step is explaining why rolling forecasts would improve business management and performance. Many people are likely well aware of the weaknesses of current processes. If this includes a traditional budget, these weaknesses include:
- Major time and energy commitment to create a budget
- Budgets are quickly obsolete
- Setting budgeting amounts is fraught with gamesmanship
- A “use it or lose it” mentality for budget allocations
- Sticking to the budget may cause missed opportunities or delayed responses to challenges
You must also prove that the benefits of rolling forecasts that I listed earlier should address some of these problems.
Top leaders are the first to convince that rolling forecasts will improve their effectiveness. Finance staff may even need to show a prototype of the reporting so that executives can understand rolling forecasts. The support and commitment of top leaders will be crucial to the implementation.
Even though managers at the company are well aware of the weakness of budgets for forecasting, change is hard. Existing processes have benefits they may fear losing. If the existing process is traditional budgeting, these benefits may carry over to rolling forecasts or are captured in new processes. Be honest if an existing benefit is diminished, but point out how additional benefits outweigh that diminishment.
Some of the biggest fears of managers when implementing rolling forecasts are less control over their destiny, new ways they will be measured in which they may not fare as well as current ways, and getting fewer resources. They have learned to play the current management game. You are now changing the rules.
Showing the weaknesses of current processes is not enough. You must also prove that rolling forecasts are a better way to manage going forward. Some things you will need to address are:
How will rolling forecasts fit into current processes?
Part of the answer is addressing what rolling forecasts replace. Rolling forecasts, along with other processes, may totally replace traditional budgeting.
If this is true, list those other processes. Be very clear about the current budgeting uses that the forecast will not replace. Point to the processes that will be replacing those uses. It’s easy for people to think of rolling forecasts as “continuous budgeting,” which it is sometimes called. Rolling forecasts often have very different functions than budgets. The more forecasts are seen as replacing all the purposes of the budget, the more the forecasts will be subject to traditional gamesmanship.
When managers ask about how rolling forecasts fit with current processes, they may really be asking if it’s just one more thing overworked managers need to do. Be honest if it is. Try hard to find something managers can quit doing to free up time for rolling forecasts.
If you’re doing a phased or parallel rollout, they will be doing forecasts and current processes for at least some period of time. Those processes may never go away if forecasting is being layered on.
Either way, you will need to provide your best estimate of the time rolling forecasts will take once implemented.
What will the rolling forecasts be used for?
One potential assumption of staff is that it will be used for the same things as budgets. Many of the answers to this question were covered in the lesson on budgets vs. forecasts and the lesson on performance management.
One of these is selecting capital expenditure investments. Rolling forecasts should not be directly used for this. Forecasts may show estimates of the company’s future capacity for capital expenditures, but the approval and queueing of these should be managed via other processes.
The impacts of these investments do need to be layered into the forecast. Build the model so they can be easily added, removed, or moved to different times.
Another assumption may be that it will be used to set staffing levels. Once again, this should not be the case. Staffing may be set via other processes like operations staffing models, traditional budgeting, or varied to target levels of sales or production.
Another assumption is that forecasts will be used to set performance targets for incentive compensation or performance reviews. Once again, these should be set in separate discussions and processes other than rolling forecasts. This is a key tenet of rolling forecasts to protect their integrity and accuracy.
Rolling forecasts may entail an aspect of performance management. Managers may be assessed on the accuracy of their forecasts. No one is perfect, but driver predictions should not be consistently wildly off. They should also not show a consistent bias above or below actuals. A manager’s job isn’t to achieve or exceed a forecast. Their job is to give their best guesses of the outcomes. This may reduce the temptation to “pack” capital expenditures or staffing.
After a Finance leader has shot down all these misconceptions about the uses of a rolling forecast, managers may be left wondering what rolling forecasts are useful for. Rolling forecasts are optimized to provide timely and accurate projections of key drivers and the financial outcomes of those drivers. It provides better cash flow and financial forecasts. Specifically, rolling forecasts are very useful for:
- Cash flow management
- Tax planning
- Capital planning
- Investor guidance
What Will the Implementation Process Look Like?
Even if managers buy into the long-term benefits, they may dread the short-term pain. All implementations take time and come with unexpected bumps in the road.
The overall process is explained later in this lesson. Some key things to communicate are:
- It will be gradually rolled out
- It will go through testing before it replaces key processes or is rolled out to more parts of the company
- Managers will have Finance assistance throughout the rollout
This means that Finance will need to:
- Build out the details of the rollout (as outlined in this lesson). Key parts of this will need to be communicated to managers.
- Schedule it into their workload to dedicate the time for a successful implementation.
- Train managers throughout the process or work with a Training department to build and lead the training.
- Be available for questions.
Identifying Parameters
Following is a list of forecast parameters that must be determined. Decisions on these parameters will determine what information needs to be gathered and who will need to be included in the implementation.
Drivers
Determining the drivers is the biggest decision in terms of forecast accuracy. In terms of implementation, start with revenue drivers and then move to cost drivers. Revenue drivers tend to impact many revenue and expense amounts.
There may be dependencies between drivers. For example, sales staff may need to set their drivers before production and operations staff can set some of their drivers. Every handoff of information from one group to another increases the forecast development time and complexity. If at all possible, see if production amounts can be set to sales drivers rather than their own drivers.
Forecast Frequency
Frequency refers to how often the forecast is updated. A popular choice is quarterly, but some companies may choose monthly reforecasts. As noted earlier, you may want to forecast more frequently if your business environment is highly uncertain or volatile. Increasing the frequency of forecasting will likely require increasing reporting efficiency. That efficiency can be gained with automation and/or reducing the model complexity, especially by reducing the amount of driver information that needs to be gathered.
You may choose to reforecast at different intervals during the implementation than once the rolling forecasts are fully in production. I refer to production as the stage when process development is done. Rolling forecasts are then periodically created and distributed according to procedures that rarely materially change.
You can decide to increase or decrease the frequency during the implementation period. You may forecast less frequently at the beginning because you anticipate that each round of forecasts will require a great deal of training, backtesting, model refinement, and report development. On the other hand, you may use a rapid iteration approach by making frequent forecasts with small and rapid changes between forecasting rounds.
Forecast Horizon
The forecast horizon refers to the length of time into the future that the forecast stretches. You may want shorter horizons if your business environment is highly uncertain. Longer horizons fit better with businesses with longer business cycles or less uncertainty.
I love wine. The wine in Washington State, where I live, is incredible. Red wine often sits in oak barrels for two years before being bottled. After bottling, wineries may hold it for months or years before selling it. That’s an incredibly long business cycle. However, wine sales, especially for premium wine, can be highly correlated to the strength of the economy. Wineries might choose to perform long forecasts due to their long business cycle but model multiple economic scenarios due to the volatility of sales. Once again, choices in one parameter are combined or balanced with choices about other parameters.
You may choose shorter forecast horizons (i.e., three to six months) during implementation. You would then lengthen the horizon in the later stages of implementation as forecast accuracy increases.
Level of Detail in the Forecast
Will forecast report rows be:
- General ledger accounts?
- Financial statement line items?
- Higher-level than financial statement line items?
Forecasts can generally be made at highly summarized levels versus detailed budgets. Push back against complexity. Ruthlessly focus on only enough drivers to provide acceptable accuracy for decisions. That level lies on a continuum of analytical accuracy and timelines, which I discussed earlier.
Organization Levels
At what organization levels will rolling forecasts be developed?
- Departments and divisions?
- Product groups?
- Company only?
Forecasts are often used for cash flow, tax, and capital expenditure planning. Different companies do this at different levels of the organization. Tax differences between states or countries could require forecasts for each of these.
Highly centralized companies with a top-down management style might tend toward only using forecasts at the company level. Companies with decentralized management processes would develop rolling forecasts at multiple organization levels.
Each organization level can choose its own forecast parameters, subject to coordination with the company-level forecasts. For example, new product groups in dynamic markets may forecast more frequently than done at the company level.
Financial Statements
What statements will be forecast?
- Income statement?
- Balance sheet
- Cash flow statement?
Many balance sheet and cash flow items may not be driver-driven. For example, investing and financing activities are management decisions with large implications for the balance sheet and cash flows. You will need to identify ways to gather these items.
Each statement will require its own variance reporting and testing. Thus, each additional statement adds complexity, but the interaction between them may require all three to be forecast. The balance sheet may drive some income statement items. When I was in banking, we budgeted balance sheets, and the net interest income section of the income statement was calculated from those balance sheets.
Gathering Data
After the parameters are set, it’s time to start gathering data. Decide the input process if you are using a “bottom-up” gathering of inputs. Here are some pros and cons of bottom-up data collection.
- Pros:
- May build ownership in forecasts: When line managers provide driver forecasts, they have a share in the accuracy of the final forecast. It’s not “Finance’s forecast” that they can completely disown.
- Captures future changes that aren’t reflected in past historical trends or relationships: Financial staff have a limited view of business operations. We look at numbers and then intuit the actual events that led to them. Managers throughout the company are much more aware of what’s happening to drive the numbers and how those drivers may trend in the future.
- Cons
- Time-consuming: I’ve mentioned earlier that process dependencies create time and complexity. It’s easy and fast for a small group of people to make a company-level forecast. The improved accuracy noted in the pros above comes at an efficiency cost.
- May be less accurate: I’ve mentioned that it’s hard to break the gamesmanship budgeting habits. Some managers may grab target or pipeline reports for forecasts, which are often overly confident for forecasting purposes. Managers should be held accountable for achieving sales targets and providing reasonably accurate forecasts. These are two different processes with two different purposes.
Creating Initial Forecasts
You now have some data with which to build a forecast model. Start with a small group of people, working with a very small number of drivers, and using a simple model.
Focus on effectiveness (i.e., reasonable accuracy), not efficiency, in the early stages of implementation. Having information that leads to better decision-making is the main criterion for the success of rolling forecasts. Efficiency can always be added later.
Widely distributed bad information will quickly kill the credibility of the forecasts, which will prevent them from being used as a decision-making tool. Mark reports as “DRAFT” if there are any questions about their reliability. Treat this as a beta process, not a final version that can be fully relied upon.
Backtesting Forecasts
Early forecasts will need significant testing and scrubbing. Finance managers will need to make sure their staff have time to commit to this. Variances to actual results can come from:
- Drivers that could have been better forecasted. Learn how to make better forecasts in the future. Sometimes, driver or model errors are more easily found with hindsight.
- Calculations from drivers to financial amounts that need to be adjusted.
- Actual results that were very difficult to predict. These happen throughout the year. Life is full of surprises. If they tend to offset, you can ignore them in the forecast. If they tend to add or decrease income, you may want to add some dollar amount for them in the forecast. When forecasting, you can’t predict exactly what causes these variances, but you know there has been a pattern to the sum of these events.
The above shows that backtesting will need to be done at the driver level and at the financial level against actual results.
Accept that some parts of the business are very difficult to forecast. If they are a large part of the company, you may need to spend more time on them each forecast cycle. If they aren’t a big part of the company, don’t waste much forecasting resources on them.
During implementation, Finance will likely lead much of the backtesting and variance analysis. There are two options for who should research and report on the reasons for the variances once rolling forecasts are in production:
- Finance reports on them all, consulting managers when needed.
- Managers report on driver variances, and Finance reports on financial calculations.
I’ve talked about holding managers accountable for their driver forecasts. This will begin once rolling forecast implementation is done and reports are in production. Let’s assume you use a rolling quarter forecast for four quarters. Will you hold them accountable for just the next quarter or all four quarters? If all four quarters, will there be different tolerance levels for variances for each quarter (for example, a low tolerance for one quarter in the future and then a high tolerance for the fourth quarter)? Better forecast accuracy can lead to better decisions.
Scenario and Sensitivity Analysis
Scenario and sensitivity analysis isn’t about what will happen “if” the forecast is incorrect. All forecasts are incorrect. No one can accurately predict the future. Leaders have been looking for ways to predict outcomes since Alexander the Great had someone examine the entrails of sacrificed animals to predict battle outcomes. Over the centuries, we’ve gone from looking at entrails to analytical navel-gazing; nothing has been completely accurate.
As Chip and Dan Heath say in their book Decisive, “The future isn’t a point, it’s a range.” Scenario and sensitivity testing show the breadth of that range of potential results. Management then assesses the implications of that range.
Modeling may also reveal incorrect calculations. Setting drivers at extremely high or low amounts can cause obviously unrealistic financial outcomes. You may find the driver calculation is only valid for a range of driver values and that another formula may need to be used at other ranges.
Embedding into Operations and Strategy
We have (finally) arrived at the ultimate purpose for all this number-crunching. Analysis informs decisions that lead to action.
The rolling forecasts should be linked to existing operations and strategy processes. These are different for every company, so it’s difficult to give many specifics on how to do this.
Companies often conduct strategic planning at least once a year. That planning may rely on the most regularly scheduled rolling forecast, or the company may decide to add an ad hoc rolling forecast cycle immediately before the strategic planning sessions.
Rolling forecasts should drive more strategic discussions throughout the year. This occurs when managers review rolling forecasts in meetings. It occurs when Finance discusses the implications of the rolling forecast report data with management teams. Reporting on a rolling 12-month forecast keeps managers looking into the future and past year ends. This constant scanning of the horizon improves readiness for the future.
The rolling forecasts should be added to current reporting. I’ll go into more detail on reporting in the next lesson. Rolling forecasts may replace reports that project results for the current fiscal year. Forecast amounts may be added to a KPI dashboard in addition to the current YTD actual and target amounts.
As rolling forecasts gain credibility, they will be used more often as a reference for other processes. Production staff may refer to it, along with sales targets and sales pipelines, when resource planning for future production. It will be used for cash flow and tax planning. And, of course, anyone in the company with incentive compensation tied to company performance will look at it to see how big of a bonus they will be getting.
For more info, check out these topics pages: