Forecasting methods

Overview

Forecasting is a key component of financial planning. However, selecting an appropriate forecasting method can often be challenging. Below, we explore seven forecasting methods, detailing their most common applications and how to implement each within Francis. From simple moving averages to assumption-based forecasting, this guide covers a range of applications to help you get started.

Basics

Year-over-year growth

Forecasting using YoY growth is relevant for companies that experience seasonal fluctuations or predictable cyclical patterns. Retail and e-commerce businesses often utilize this method.

Method

Reference revenue from the same month last year and add a growth rate to account for expected changes in performance.

Applications

Typically used for revenue projections and employed by companies who experience predictable seasonality.

$$= \text{"Revenue"}[-12] \space * \space (1+5\%)$$

Percentage of revenue

When companies scale, certain expenses and balance sheet items tend to scale proportionally with revenue. With the percentage of revenue method, certain line items are forecasted proportionally to forecasted revenue.

Method

Determine items as % of revenue based on past performance and multiply this percentage to get your forecast values.

Application

Often used for costs that scale and correlate with revenue, including working capital items such as inventory or receivables.

$$= \text{"Revenue"}[0] \space * \space \text{"COGS \% of Revenue"}[0]$$

Moving average

Moving averages are suited for companies with steady operations and relatively low volatility. This method is based on the notion that the near-term future will likely mirror the recent past, where it can be helpful to smoothen short-term fluctuations and focus on longer-term trends.

Method

Calculate a moving average based on the last X months actuals. Optionally, add a growth assumption to account for expected changes.

Application

Moving averages are used across a wide range of financial items including revenue, expenses and balance sheet items.

=\text{AVG_LAST(“Revenue”, 3)}*(1+5\%)

Annual inputs

The annual inputs method is optimal when annual costs can be estimated with high certainty, yet the monthly distribution remains uncertain. It suits companies that adhere strictly to annual budgeting and planning cycles.

Method

Input your annual assumptions and divide by twelve to distribute costs equally throughout the year, thereby generating your monthly forecasts.

Application

Ideal for fixed cost items where the annual cost can be estimated with high certainty, but monthly distribution is unknown.

$$=-10.000/12$$

Prior months balance

The prior month's balance method assumes a steady state from one month to the next and is a simple yet effective method for stable accounts on the balance sheet.

Method

Based on the assumption that conditions remain stable. Simply reference last month's value to keep items constant.

Application

Typically used for balance sheet items that have limited predictability, or no expected changes during the forecast period.

$$=\text{"Rent"}[-1]$$

Cost per employee

The cost-per-employee method is particularly relevant for high-growth companies where employee-related benefits and costs constitute most of the cost base. This is often the case in software companies, where most expenses relate to employees.

Method

Based on historical data, determine an average cost per employee and multiply by projected total employees.

Application

Relevant for cost items that scale with the number of employees, including items such lunch, insurance and office expenses.

$$=\text{"Lunch, cost per employee"[0]} \space * \space \text{"Total employees"}[0]$$

Fixed assumptions

The fixed assumptions method is particularly useful for fixed-cost items or in situations where the environment is stable and historical patterns are expected to continue into the future without significant changes.

Method

Instead of formula based forecasting, simply hardcode scheduled expenses and other forecast assumptions directly into your model.

Application

Hardcoded forecast assumptions are especially relevant for fixed-cost items including rent, insurance or interests.

$$=-10.000$$