Forecasting at category level rather than at account code level

By David Parmenter

It is a myth that we needed forecast at account code level. A forecast is a view of the future. It will never, can never, be right. “It is better to be nearly right than precisely wrong” John Maynard Keynes.

Looking at detail does not help you see the future better; in fact, I would argue it screens you from the obvious.

COUNTING THE TREES IN A FOREST

Imagine that you have been asked to count the trees in a state forest that consists of 100 square miles of trees. You have two choices, the detailed way and the “helicopter” way.

For the detailed way, you could set up 10 teams of seven people. Each team is assigned 10 square miles and is given satellite navigation equipment, a different colour of spray paint, safety gear, camping equipment and provisions for three weeks or so. The teams update their count each night on a spreadsheet. At the end, the counts are consolidated, and some data is left out because the counters in some teams forgot to load all their spreadsheets into the workbook. The final count, therefore, is wrong, although no one knows that.

For the helicopter way, satellite imaging is used to select five sample areas that are 1/1000 of the forest. The staff are assigned to five bigger teams, and each counts their area in a day. The count of the five areas is averaged and then multiplied by 1000. The answer is wrong. But it was wrong quickly and is still a good approximation. For forecasting, the helicopter way is usually the better option unless you are forecasting payroll where managers can forecast by their staffs’ actual salaries.

While precision is paramount when building a bridge, when every small detail needs to be right, a forecast should concentrate on the key drivers and large numbers.

Following this logic, it is now clear that as accountants we never needed to set budgets at account code level. We simply have done it because we did it the previous year. You can control costs at an account code level by monitoring trend analysis of actual costs over 15- 18 months.

We therefore apply Pareto’s 80/20 principle and establish a category heading that includes a number of general ledger codes .

Helpful rules that can be used to assist you apply this foundation stone:

  • Limit the categories to which budget holders must forecast to no more than 12.
  • Select the categories that can be automated and provide these numbers.
  • Separate out a forecasting line, in the model, if the category is over 10 percent of total (e.g., show revenue line if revenue category is over 10 percent of total revenue. If an account code is under 10 percent, amalgamate it with others until you get it over 10 percent. Remember that a category will have a number of account codes in it. This rule applies at budget holder and consolidated forecasting levels.
  • Map the general ledger account codes to these categories. A planning tool can easily cope with this issue without the need for a revisit of the chart of accounts; see Exhibit 3.1 for an example of this mapping.

When forecasting revenue, it is important to remember the story of counting the trees. The wrong way is to set up a schedule of every product and get each branch to fill out the details.

In one workshop, I ran for a service sector organisation the group came up with the decision that there would be a maximum of 15 categories 7 of which would be automated.

Budget holders forecast only these categories

  1.           Revenue (Three to four categories)
  2.           Salaries and wages ordinary time
  3.           Other personnel expenditure
  4.           Health and welfare
  5.           Training and conference, including travel
  6.           Operational equipment – that is not capitalized
  7.           Property costs
  8.           Promotional activities

The categories that can be automated

  1.           Operational equipment repairs and maintenance
  2.           Office equipment, computers, and consumables
  3.           Communications costs
  4.           Fleet costs
  5.           Building maintenance
  6.           Miscellaneous costs
  7.           Depreciation

For more details access my toolkit that currently is on sale:

Forecasting at category level is discussed in great length in my Toolkit How to Implement Quarterly Rolling Forecasting and Quarterly Rolling Planning – and get it right first time // Toolkit (Whitepaper + e-templates)

You can have a look inside the toolkit

Also read:

8 steps to implementing rolling forecasts

https://fpa-trends.com/article/best-practices-implementing-rolling-forecast/