Sit an advocate of plain English in front of Bloomberg TV and, within just a few minutes, I’m certain they’d be scratching their heads at the oft-used obscure jargon. For instance one evening last week a presenter said that, “Some analysts are now ‘decrementing’ a company’s stock.” To my surprise, decrement is a verb and what the presenter said was grammatically correct, but I’d never heard it before. Why not simply say “reducing ” or even “marking down”? The software industry tends to manipulate language to suit its own needs too, with for instance the mega-vendors re-labelling their Business Intelligence acquisitions as “Analytics” soon after the ink has dried on the deal. Given that not-for-profit organizations tend to shy away from anything that mentions commerce, dropping the word business was quite a smart move that should help analytic tools become more pervasive in government, healthcare and charities. But standard BI tools are a rapidly maturing segment of today’s analytics market where ‘predictive analytics’ has become all the buzz with the mega-vendors either rushing their own versions to market, mopping up any remaining independent specialists, or sometimes both – the latter no doubt resulting in some long-drawn out meetings in corporate PR departments about how to position such a two-pronged strategy.
Is it “Historical” Analytics or Predictive Analytics?
If they felt the need for yet another rush of blood to the head, our Plain English boffin should take a look at how the software industry uses the word ‘predictive’. While it is undoubtedly the right word to use when talking about applying statistical approaches to forecasting, it is now liberally tagged to uses that are anything but predictive. Consumer purchasing patterns is widely cited as a successful use of predictive analytics but when Amazon or indeed any other online retailer says, ‘We have suggestions for you’ their suggestions are based on the purchases of other consumers who have bought something identical to you at some time in the past. That’s historical analytics rather than predictive analytics, and anyone who has made occasional purchases of gifts for toddlers in the family knows what a random assortment of suggestions can subsequently be pitched to you.
That it might actually unlock some new consumer insight that leads to previously undreamt of revenue streams that will be the saviour of companies battling with the current economic uncertainty is undoubtedly one of the reasons for the sudden resurgence in the popularity of “predictive” analytics. Good luck to them; I’d give it a go for sure.
Reactionary is not “Predictive”
Having invested in the capabilities, the mega-vendors are now telling us that we need to apply predictive techniques to all lines of business. In fact, just last week a paper on ‘Predictive Finance’ appeared in my inbox. One of the suggested use cases for adopting predictive analytics in Finance is in receivables management to help detect sudden changes in a company’s payments. This too is ‘after the fact’ in that it will alert you that Company A failed to send a payment yesterday (like they normally do every Thursday), rather than actually using any predictive capabilities so that we knew well before Thursday came around that no payment would be forthcoming. And given that the sudden hiccup in Company A’s payment pattern is most likely based on some unexpected event – such as their loss of a key customer – the best foresight would typically be some qualitative field intelligence that their key account was said to be looking at other suppliers.
The result of applying Predictive Techniques
I hold similar reservations about applying predictive techniques to forecasting. It’s typically not the trended ‘business as usual’ revenues and expenses that actually lend themselves to statistical techniques that are difficult to forecast. It’s the new business initiatives where there is no history, and the anticipated – but as yet uncertain – events that may impact revenue growth in the coming year, such as a rapidly emerging new technology. Because of such issues, applying sophisticated predictive techniques on top of traditional ‘roll-up; roll-down’ planning and budgeting methodologies is unlikely to improve the accuracy of forecasts. It does nothing to improve corporate agility which is the essential capability needed to cope with uncertainty in these times. What’s more, budgets that only contain financial line items – even when forecast using the most sophisticated algorithms – provide little insight into how real-time changes in the business operation, the external environment, the customer base, or resource costs will impact financial results in the future.
No, applying predictive techniques to traditional budgeting is simply, ‘painting lipstick on the pig’. The key to improving planning and budgeting is to slaughter the pig and replace it with something that will actually ‘save your bacon’ such as a driver-based approach where business users can share both their quantitative and qualitative knowledge about the business in a model that can be rapidly re-forecast to give better insight into the future and what needs to be done to make the best out of it. That’s exactly what you get with Anaplan.