Data Driven Decision Making

Fiscal Accountability The use of budget data for decision-making is a garbage in, garbage out scenario. If the assumptions built into the budget are poor, then the decisions made will be poor. As such, it is important that budgets are constructed with the best information and analysis available. There are different types of budgets, so the first part of the process will be to select the best type.

In many organizations, static budgets are the norm, where the organization sets out a budget based on incremental changes to the previous year's budget. This technique has the benefit of being easy to implement, but it is not necessarily going to yield the most robust data.

If the organization is exceptionally stable and has many years of historical data from which to draw its estimates of change, then the budget may well be accurate, but for many organizations this is a poor way to get an accurate budget, and decisions made from such a budget will also be poor. Managers use this data for many types of decisions, including strategy formulation, resource deployment, and tactical decisions. The budget might show that expected sales of a product are down, and inventories will increase.

The manager might respond to this projection by ordering fewer items produced of that product. A manager might lay off an employee if a division looks to lose money in the coming year. So there are a wide range of strategic, tactical and operational decisions that are made on the basis of budget data, which highlights why that budget needs to be as strong as possible. Data-driven decision making is a process by which budgets are set on the basis of statistical analysis.

A flexible budget is one that starts with a static budget, but has flexibility in that the figures are adjusted with each quarterly period (or more rapidly) to reflect recent changes. This is one form of data-driven budget process, but there are others. Some budgets benefit from adjustments to forecasts based on qualitative analysis, and in other cases quantitative. The key is that new variables are introduced.

If a static budget is created based on past budget (or even past performance) and projected growth rate (usually historical growth rate), a data-driven approach would seek to introduce the influence of other variables. These could be any variable that has been tested and deemed to be relevant. Maybe the interest rate is relevant, and it is expected to change in the next year. The budget would incorporate that change.

Data-driven decision making requires having the measures in the first place, but then it also requires testing variables for their fit, and then using that data to derive equations that help managers create forecasts that are more likely to be accurate, within a tighter confidence interval. This starts with gathering more data points that can be identified as variables influencing different parts of the budget (ALCTS, 2016).

There are still challenges -- there are costs associated with gathering the needed data and then giving it the needed statistical analysis, but if these problems can be overcome, the budget is more likely to be accurate (Chen, Lin and Zhou, 2015). The analysis used to derive better budgets should be able to identify meta-patterns and outliers as well. An example would be advertising revenue at a television station.

Perhaps that revenue spikes every four years because of the Olympics -- that is something that analysis needs to pick up to deliver a better budget. Ultimately, decisions have to be made on the basis of the budget; that is the point of the budget. The decisions therefore require the best budget possible, which is where statistical analysis enters in. Having different scenarios presented is also important, because that allows for more rapid decision-making should the "worst case" or "best case" scenarios be coming to pass.

Again, a regression analysis can help with this. For example, maybe a company's worst case.