Time series analysis is all about understanding the past to make better predictions about the future. A company that uses time series analysis for forecasting will collect specific data at regular intervals over a period to reveal underlying patterns and trends.
Time series analysis can be used in a variety of ways, including forecasting cash flow, sales, or production levels. Common applications of time series analysis in business include:
- Budget forecasting
- Sales forecasting
- Stock price analysis
- Inventory analysis
- Workload projections
- Production forecasting
- Yield projections
Finding and capturing data shouldn't be an issue for many businesses, says Matt Rogers, Executive Vice-President EMEA at data forecasting company OneStream. “There is a huge amount of data in most businesses. The challenge is understanding what to capture and how to interpret that data,” he says.
The first step is to decide on the right information to track, says Rodgers. “You’ll have sales, operations, HR and a whole host of data sources, but which of these data adds value to your strategic goals? That’s the data you need to isolate and track,” he says.
Once you've collated data for the period you want to assess - be that a month, year, or several years - you can then begin your analysis:
What are the four components of time series analysis?
Trend
The trend component refers to the long-time movement or path of data over time. Trends help business owners identify whether data is generally increasing, decreasing, or remaining constant. Therefore, trends can provide a fundamental picture of growth, decline, or stagnation.
Seasonality
The seasonality component refers to data fluctuations that occur at fixed intervals within a specific period. It is often tied with calendar events like holidays and seasons, and it helps business owners predict things like revenue and sales for the year ahead.
Cyclical variations
The cyclical variation component refers to medium-to-long-term data fluctuations over some time. However, this component differs from seasonality because it doesn't have fixed reoccurring intervals. Cyclical variations are also often linked with political, economic, or social events.
Irregular fluctuations
The irregular fluctuations component refers to unpredictable or seemingly random variations in the data that cannot be explained by trends, seasonality, or cyclical variations. Also known as residuals or anomalies, this component is important because identifying it too often may point to data inaccuracies.
Why and how to use time series analysis to help your business
Time series analysis is useful for tracking things that can vary over time, like sales, employee turnover, productivity, or pricing.
Measurable.energy makes AI-powered electrical sockets that can reduce energy consumption. The company has put forecasting at the heart of its business since 2018, explains Dan Williams, CEO.
"Our entire business is built on forecasts of things like energy prices, and what targets will be in place around things like corporate energy consumption,” he says. “So we’d track how energy costs and energy consumption change month to month, year to year, and that allows us to forecast what might happen not just for the next year, but more long-term, over the next ten years.”
You can run time series analysis using most commercial software tools, or an Excel spreadsheet. In some cases, businesses will work with industry analysts, market researchers or specialist consultants to buy in prepared forecasts.
“Businesses that are in a growth phase may not have the internal resources to manage even their data, much less the 80% of data that comes from outside their organisation, like government statistics around things like inflation or employment rates,” adds Rodgers.
Time series analysis example
Measurable.energy works with several specialist firms to secure forecasts for energy prices, rather than tracking this information internally. Global conflicts and the pandemic have made energy prices rise rapidly, meaning the business now works with specialist analysts who produce time series forecasts based on shorter periods, explains Williams.
“Once upon a time, you could forecast energy prices based on 20 years of historical data, but now it’s more like three years,” he says. “It has become a very challenging area to forecast, and so we work with specialists who can track things very closely.”
Because time series analysis is a quantitative model of forecasting, it provides business leaders with statistics that can be used to create budgets, identify funding needs, and create timelines for future business projects. If the underlying data for time series analysis is collected regularly, over several years, then it can provide extremely accurate projections.
“We rely on forecasting data to help us understand things like marketing demand and pricing,” says Williams. “If we know that energy prices are likely to increase and that regulations are coming into force around carbon emissions, then we can use that to help predict our sales, but also inform things like recruitment and marketing.”
Advantages and disadvantages of time series analysis
Advantages of time series analysis
The key advantage of time series forecasting is that it provides businesses with a clear, statistical view of where the business stands. By putting business activities into a standardised data form over time, you can ignore a lot of ‘noise’ and get a clear picture of what’s happening over time.
The statistical models involved in forecasting can also help to uncover patterns and relationships that might not have been immediately apparent. For example, a business might discover that an increase in one cost is correlated to an increase in production output, or profit.
Accurate forecasting aids your business in managing cash flow, and identifying when additional funds are required for operations. With up to 54-day payment terms, the American Express® Business Platinum Card helps give you greater cash flow management flexibility¹. Moreover, Amex Business Platinum Cardholders receive an annual £150 to use at Dell Technologies, supporting investment in cutting-edge tech for business analysis and forecasting².
Disadvantages of time series analysis
The key disadvantage of time series analysis is that the quality of the forecast will always depend on the quality of the data. If your data doesn’t span a long enough period, or it wasn’t collated well enough, then the insights may be less conclusive.
In general, it’s wise to use data that has been captured at very regular intervals of three to six months, for at least three years for this type of forecasting. However, in some cases, even the experts get the forecasts wrong. “So for businesses, it’s never going to be a perfect science,” says Williams.
1. The maximum payment period on purchases is 54 calendar days and is obtained only if you spend on the first day of the new statement period and repay the balance in full on the due date. If you'd prefer a Card with no annual fee, rewards or other features, an alternative option is available – the Business Basic Card.
2. Once you have enrolled your Card Account, you’ll get up to £75 in statement credits between January and June, and the same between July and December for United Kingdom purchases with Dell Technologies on your Business Platinum Card – up to £150 in statement credits annually. Terms and conditions apply.