Sales forecasts help you budget and manage your business. They provide a road-map for your business. Changes that occur can be managed effectively and strategic adjustments can be made for what is working or not working. The time series method of forecasting uses a model to predict future values based on previously observed values. By using the historical data we already have, we can predict future sales.
Time series is a series of data points listed in chronological order, taken at successive equally spaced points in time. For a company that sells products to consumers, these data points are often taken daily or weekly. For other types of businesses, the data may be measured hourly or monthly. The time series method of forecasting is more reliable when there is a large set of data that is representative of the category – both in the number of similar items being observed and the amount of time (for example, sales on this particular date for the past three years) being observed. Using a large sample enables the random variations and anomalies to be “smoothed” out, providing a clearer picture of the pattern of selling behavior.
Once the data points are collected, they can be plotted on a sales curve. Each point on a sales curve is assigned a mathematical value. For example, a consumer goods retailer may use a weekly sales curve, where each week of the year is assigned a percentage based on how that particular group of items sells relative to the other 51 weeks of the year. The total percentage for the year is 100%. The sales curve will take into account seasonal variations that occur with this category. For example, winter coats will have a different sales curve than shorts and tank tops. Other factors that can affect sales curves are promotional activity, trends and cycles. Items within a company may have different selling patterns, as in the winter coat and shorts example. Because of this, a company may utilize many different sales curves. The key is to assign all items to the sales curve that is most representative of that item’s selling behavior.
Once all items have been assigned a sales curve, a company can use current demand to predict future demand based on the mathematical function of the sales curve. For example, if we sold 80 units in the past 8 weeks and we know that the past 8 weeks represent 20 % of the year’s selling on that item’s respective curve, we can forecast that we will sell a total of 400 for the year (80 / 20% = 400). If the next 12 weeks represents 40% of the curve, we can predict that we will sell 160 units in the next 12 weeks (40% x 400 = 160). Of course, this is a very simple example. You will want to remove any anomalies in the sales data that will most likely not repeat themselves and factor in things like inventory availability and changes that may be happening in the marketplace. Human interaction is required!
By using the time series method of forecasting, companies have a reliable method of forecasting future sales. Good forecasting leads to improvement in operational performance. Inventory levels can be adjusted, leading to improved turnover and cash flow. Pricing can be adjusted to spur demand or increase profitability. Strategic decisions can be made more quickly utilizing the roadmap that the forecasts provide.