Sat, Aug

Pondering PREDICTIVE Analytics?

Walter Hale considers whether this business modelling method has a place within PSPs.

If anyone remembers American economist Irving Fisher now it is because of his unfortunate remark, in late 1929, that “stock prices have reached what looks like a permanently high plateau”. Months later, the Wall Street Crash proved him spectacularly wrong and cost him eight to ten million dollars personally. It was this kind of prediction that provoked John Kenneth Galbraith to famously remark that, “the only function of economic forecasting is to make astrology look respectable.” 

The urge to know what will happen next - and put that knowledge to use - is as old as the human race. And today, eightysomething years after Fisher was proved so disastrously and expensively wrong, businesses are being asked to invest in something that has a reassuring amount of science, software and technology behind it, a technique known as predictive analytics. In essence, this is a form of business modelling that can help companies anticipate what is coming, what is changing and what is different in one context or another - and do so more accurately than a bunch of managers drawing on their perception of the past (which may or may not be entirely accurate), experience (again highly subjective), personal contacts and gut instinct to predict the future.

The range of applications for predictive analytics is intriguingly diverse. Ikea is experimenting with this modelling to reduce checkout queues at peak times. By analysing warranty claims more deeply, PC maker Lenovo cut the costs of such claims by 10-15%. Basketball team Orlando Magic has even used predictive analytics to decide its starting line-ups, with some success.

These are all relatively large organisations and many owners of SMEs, such as those that predominate in the wide-format printing industry, have traditionally taken the same line as they have with big data: that this is all just too complex and costly for them, that it won’t dovetail with their existing, perhaps slightly patchy, corporate systems - and, on top of all that, is of little relevance to companies in the B2B sector.

There is some truth in all these objections. Some of the companies selling these solutions - such as IBM - are not selling them cheap. The systems are only as good as the data that is put in, and what boss of a small business has the time or resource to collect relevant and useful data? Some of the vendors can make the proposition sound more cumbersome than it ought to be - for example, SAP’s daunting pitch to SMEs calls on them to “embed predictive analytics into all areas of an organisation, from point of sales to the call centre”. And it is fair to say that, until now, predictive analytics has proved particularly valuable by, for example, analysing trends on social media to help brands and retailers understand and anticipate consumer sentiment and react accordingly.

Yet that isn’t the whole story. For a start, one of the most significant business trends of the last few years is the ‘consumerisation’ of B2B, driven by the realisation that companies aren’t selling their services to other companies, but to people within those companies. So the techniques that have helped the likes of Ikea understand its customers are increasingly likely to help you understand yours. The old fashioned sales-rep driven sell is giving way to a process where a customer’s preferences are already influenced by communities, campaigns and online information. Your ability to track those communities, campaigns and sources of information may help you close the deal. It’s worth remembering too that with the advent of web to print, many print service providers maybe dealing directly with consumers some of the time.

The boom in Cloud-based analytics, capabilities and services has made it much easier for every organisation to develop a reasonably sophisticated analytics capability without becoming internal experts or betting the farm on this technology. Platforms like Microsoft Azure make such capabilities available to organisations of every size. And much of the data you need already exists - in your management information system even though, in one supplier’s estimate, only a third of wide-format printers in the UK are actually making full use of these systems.

And although everything that is digitised can be quantified - and everything that can be quantified can be measured - you don’t have to analyse everything. Rob Symes, CEO of The Outside View, a data consultancy focused on the property sector, says: “The often limited ability of SMEs to define the critical business questions that is the biggest barrier to implementation. It’s one thing to measure the weather, it’s another thing to see the snow in the summer. You need to measure what’s meaningful - meaningful to you.”

Symes argues that SMEs have one significant advantage over big players such as Amazon and Tesco that are renowned for their use of data. These companies have to get to grips with an enormous, anonymous, customer base. A PSP can focus much more narrowly - you have fewer customers and know who they are.

You could also decide that it’s more important to use the data your business generates to look at a specific project, model various scenarios and share the results with employees, hopefully making them more engaged and effective. A very simple use of this technique would be to create heatmaps to track mouse movements across your website - it could give you a usefully different insight to the usual Google Analytics offering that focuses on traffic, visits and visitors.

The potential gains with predictive analytics are hardly insignificant. If you could develop models that helped you forecast sales more accurately, control your inventory more efficiently, market more effectively, smooth your cashflow, prepare your business for dips and spikes in demand and feed the insights you have acquired into your business plan, why wouldn’t you do that?

The key here is to decide what your most important goals are, and don’t aim to do too many things at once because too much data can bewilder, befuddle and becalm. You may, for example, decide that the best thing you can do is harness the power of predictive analytics to profile your customers, identify the characteristics that your best customers have in common and then target your marketing to find more like them. Or you may, if you have suffered a churn of employees, want to delve into the detail and identify likely leavers.

You don’t have to invest in all-singing, all-dancing, bleeding edge of technology solution. You could start the journey by buying in a desktop package like Microsoft Azure and start experimenting. You could also canvas the views of your management information system supplier. You could even, as Symes suggests (albeit with a certain amount of self-interest) outsource predictive analytics to a consultancy as a low risk, relatively inexpensive way of beginning the learning process and staying agile as the technology and software continues to develop.

Sean Price, head of performance at business intelligence company Datanostic, says: “There are some very disruptive emerging low-cost technologies that, when combined with the talent, experience and wisdom, makes solutions affordable and accessible to small and medium sized businesses, closing the gap on larger technology-enabled companies. There are companies that can help you ask the right questions, develop technology-based solutions and provide the answers on phone, tablet or desktop, whatever way you want it. These solutions can be low-risk, cost effective and subscription based - just cancel if they’re not delivering the return on investment.”

Whether you develop the resource in-house - or outsource it as Symes and Price suggest - you don’t need to get bogged down in the technological intricacies and complicated mathematics of data science to make effective use of predictive analytics. The key is not to let the technology distract you from your business goals.

Used effectively, predictive analytics could you help your business make better decisions. A little objective data could help challenge conventional wisdom and the inward-looking groupthink that can afflict any group of managers who have worked together for a long time.

And every print service provider - no matter how smart management might think they are - could gain from making better decisions. You don’t want to be the next Irving Fisher, do you?

Upcoming Events