Predictive analytics: how small improvements can deliver big results

Results directions. Choice for easy way or hard way.

It never ceases to surprise me at the wide array of interesting and smart folks we have the privilege to meet through the course of our work at Smart Vision Europe. We are two months into 2014 and I have been checking back on my diary to make sure there are no loose ends. I’m encouraged by how busy we are and by the diverse range of businesses we’re talking to – so much so that I thought it was worth sharing. The purpose of sharing is not to showboat about how busy we are but to reinforce the point that for almost every type of business or organisation, above a certain size and complexity, there are almost certainly two or three areas where predictive analytics can help improve performance.

Conversations we’ve been involved in so far in 2014 include:

  • A major supplier of security technology looking to provide predictive fraud, risk and non-compliance detection for their financial services and central government customers.
  • A major energy distribution business with a range of analytics applications in mind from predictive asset management and quality assurance through to predicting the likelihood of apprentice engineers completing the development programme above the required standard.
  • A global cellular network infrastructure business looking to optimise maintenance interventions and schedules to minimise service disruption and outage times for its clients
  • A multi-channel retailer grappling with an array of potential opportunities to improve customer life time value, customer retention and marketing optimisation across multiple channels.
  • A European hospitality business who knows that predictive analytics will help it improve the occupancy rates at its properties whilst improving profitability through more effective guest insight, understanding, prediction and marketing intervention.
  • A premier, multi brand recruitment and HR business with a 2014 strategy that explicitly includes the use of predictive analytics to enhance the profitable placement of permanent and temporary candidates by better predictive matching of candidates to opportunities.
  • A regional social housing provider using predictive analytics to understand key drivers of tenant satisfaction in order to minimise rental arrears (dissatisfied tenants are much more likely to withhold rental payments as a means of peaceful protest).
  • A large UK leisure club business that sees real opportunity to improve its business performance by predicting which members are most likely to cancel their subscription early enough in the cycle so they can intervene and bring about a better outcome.
  • A UK-based general insurance business that has already seen strong returns from its investment in predictive profiling of prospective customers and now wants to scale up its activity and enable more automated deployment of the predictive analytics outcomes.

Something I have noticed about all of these applications is this:  in every case a performance improvement of fairly modest proportions – say a 3% improvement in member retention, a 5% lift in average transaction value, an 8% reduction in unexpected service outage or a 1.5% improvement in profitable placements – will make a world of difference to an organisation’s performance and competitive advantage. Incremental improvements that might seem small when viewed as percentages can actually equate to many thousands pounds onto the bottom line.

As I said earlier, I firmly believe that almost every organisation above a certain size can benefit from predictive analytics, usually in several different ways. So what kinds of organisations are those? I’ll stick my neck out and share some general rules of thumb.

  • Any B2C or B2B business (or charitable organisation) with more than about 30,000 clients (supporters / users / visitors / assets) that can be identified and where some data is held about them.
  • Any organisation which has interdependencies between multiple processes and procedures and where data can be collected on the performance of those processes or the behaviour of customers or users within those processes.
  • Any organisation concerned about operational efficiency and / or profitability.

You do need some data in order to do predictive analytics but it’s a mistake to assume that your data needs to be clearly structured and organised in advance. In every organisation we work with the data is always sub optimal, it is never all in the same place or neatly organised and it has rarely been collected with analytics in mind.  Yet the reality is that when we start to explore that data it almost always turns out to be more usable and valuable than the client ever imagined.

If your organisation is not yet considering the application of predictive analytics, why not?