Customers, eh? Can’t live with them, can’t live without them. I mean, you might literally not be able to live without them, what with them paying your wages and all that. If only they were a little more predictable, though...
We won’t need to tell you that being able to accurately forecast demand for your products among customers is a rather valuable skill. Not only does it allow you to plan ahead, it also means you can optimise the amount of stock that you keep. This, in turn, helps to minimise the chance of stock running short and of you missing orders, or the possibility of buying more stock than is required and having an unnecessary amount of capital tied up in it.
Adidas’ work with big data is just one example of just what an impact accurate forecasting can have for a business. The sportwear brand grew its online sales from zero to over €1 billion between 2010 and 2016, using a combination of automation and data crunching. Such automation is commonplace nowadays, with firms routinely using automatically triggered emails and the like to reach out to customers. Adidas used its data, though, to forecast what would motivate and drive purchases in consumers and then inform the automated activity that was carried out.
There’s no one right way to forecast demand, with a variety of different approaches able to be employed. These can typically be grouped as survey methods (those via which people are polled about their buying intentions) and statistical methods (those via which data is examined to infer likely future demand). The method an organisation chooses to use can simply be down to which works best for them.
It’s for this reason that Peak looks at a variety of different models* for our demand forecasting clients, before choosing the one that is best suited. We’d argue that any company offering demand forecasting should do the same.
Where our approach differs to that of most other companies offering demand forecasting is our ability to not only uncover insights about your business, but to recommend subsequent actions to take, too.
How? Well, we view ourselves as being in the business of business improvement, rather than just data analysis. There’s no point in analysing an organisation’s data out of context, so we build a deep understanding of each of our clients before we do anything else, including their current situation, their processes and their goals. In this way, our analysts can deliver tailored business recommendations in plain English, and we can even feed actions directly into an organisation’s existing IT systems.
*For those of you who are interested, we typically use a company’s sales volume data with macro and micro economic data and our forecasts use techniques that include distribution fitting using seasonal ARIMA models, exponential smoothing and Bayesian inference. We use the results produced from these methods alongside mixed integer programming to determine optimum product stocking levels.