Data-driven innovation

Still in the early adopter phase, some shining examples of predictive analytics demonstrate the potentially enormous benefits it can deliver.

A strong focus on data management enables organisations to ensure data flows are efficient and transparent.

It can also be used for predictive data analytics to improve the speed of decision-making.

The recent explosive surge in big data means predictive analytics is gaining wider acceptance for operational use.

James Fisher, vice-president of product marketing analytics at SAP, cites IDC figures of an advanced analytics market worth $3bn by 2016. He claims many SAP customers use Hana, the company’s predictive analytics platform, to drive profit and cut costs by applying the software to analyse their data in ways never previously thought possible.

Within the context of customer relationship management (CRM), however, predictive analytics is more at the experimental stage rather than a de facto modus operandi. That said, it is no surprise that companies whose business model is built on technology are in the vanguard when it comes to applying predictive analytics to CRM.

A tactical game

Gaming company Bigpoint, the creator of Battlestar Galactica, is using predictive analytics to monetise players and increase revenue by a projected 10% and 30% a year. Battlestar Galactica has nine million registered players and notches up 5,000 events per second. Bigpoint’s predictive model is allowing it to intelligently make real-time decisions about a player’s actions. For example, if a player’s ship is destroyed, it acts as trigger for the predictive engine to analyse previous gaming behaviour. If appropriate, a personalised context-related message offers the player a new ship – for a small fee, of course.

Bigpoint is not there yet, but it is confident of growing annual revenues.

Adept at tax collection

Her Majesty’s Revenue & Customs (HMRC) has created a predictive analytics platform to improve debt collection and risk evaluation. Its platform, called Adept, integrates analytics into debt management and is designed to customise debt collection interventions for millions of late tax payments each year.

Adept is integrated with HMRC’s collection systems and uses predictive modelling to inform more sophisticated risk and behaviour-based collection strategies. It identifies different types of debtor groups and targets its communications to these groups based on their specific attributes. To date, it has been so successful that HMRC estimates it will collect an additional £3bn of debt by March 2015.

The system uses “behavioural economics” predictive modelling to reveal whether SMS, landline, mobile phone or printed letter is the most effective channel for communication. It also assesses the message content, or trigger. By mentioning the public services funded by taxes in one letter to late payers, for example, the payment rate increased by 20%.

HMRC is also learning from the financial servicesindustry and using predictive modelling to identify customer behaviour that provides early warning signs of default on arrangements.

The question of data accuracy is an important one, however. If the data is skewed with errors, the models are going to be inaccurate. To ensure that data is accurate, HMRC conducts a quality check as data is shipped from the debt collection system into Adept and converted into analytical form.

The result is that HMRC has a flexible, easy-to-change approach to mass customisation, enabling it to assign the most appropriate sequence of collection actions to each debt. Events from a customer promising to pay or missing a payment deadline are used to trigger automatic re-evaluation of past decisions. The business actions cover the full range of debt collection interventions, including letters, phone calls, visits by field force agents, referral to a debt collection agency and, ultimately, court proceedings.

The groundwork for Adept was laid several years ago, when debt management systems were integrated to create a single system which provided the flexibility to create workflows that generate letters, drive predictive diallers in contact centres, manage door-to-door collections and carry out legal proceedings.

HMRC created an analytical prototype on a standalone system with a monthly data feed from the integrated debt management system. It also created a new support model and technical mechanisms for sharing ownership of different parts of Adept between IT and the business. Creating this system involved integrating four different Oracle databases and four different SAS tools into a seamless analytical environment. The most complex phase delivered a 12.5TB analytical database processing millions of new records each day, generating event triggers for automated decision-making and updating analytical models based on billions of records.

Through Adept, HMRC is transforming the way it uses data. In fact, more data sources are being added, so new connections can be made. Its data-gathering powers means it is able to gather bulk data about businesses accepting credit and debit cards. By using a big data system called Connect, provided by Detica, HMRC says it is able to cross-match more than one billion pieces of data to detect risky taxpayers to investigate.

Analysing data for retail rewards

In the retail sector, a number of companies are moving forward with predictive analytics. One example is Tesco, which is building on its reputation for employing new technologies to enhance CRM. Alys Woodward, research director at IDC, points to a scheme Tesco has initiated which offers customers, in real time, products related to existing purchases. A loyalty card is inserted into a device that is fixed to a shopping trolley, through which tailored offers are made to the shopper as they place items in their trolley. For example, a shopper may place a barbecue chicken in the trolley and then receive a 3-for-2 offer related to the item, such as charcoal for barbecues.

By recognising who the customer is and the purchases they have made, Tesco is attempting to predict what they are going to do next and intervene with relevant offers. It is a simple premise – and one that, if successful, could attract customers and lead to greater profits.

Clearly the prize is great, and to illustrate just how much some retailers are investing in this area Wal-Mart recently acquired predictive analytics firm Inkiru. The US retail giant says it will now accelerate big data capabilities, such as website personalisation, fraud prevention and marketing.

Richard Kellett, UK marketing director at SAS, is anticipating a far greater upsurge in predictive analytics. “Knowledge and awareness is increasing and there is a greater openness to seeing predictive analytics as an operational tool rather than just a strategic tool,” he says. “[In the UK] we are lagging behind other countries, but there are pockets of excellence.”

He cites Waitrose as one example. The company is applying predictive analytics in its supply chain to ensure it cuts down on waste and does not under-supply, so customers are not walking away to a rival. In short, it is mapping sales of individual stores and then factoring in components such as weather maps and forecasts to determine shipping quantities, locations and timings.

Life-saving potential

The financial services industry is using predictive analytics to refine its debt collection methods. Some companies are recognising that a certain type of customer responds negatively to payment reminders – if left alone they will make the payment, but if nudged they will dig their heels in and delay payment. By identifying these behavioural trends, companies are able to optimise their operations and improve customer relations. At a wider level, predictive analytics is making inroads into many areas of life. For example, a trucking company in the US is using the technology to predict fatigue levels in drivers, cutting serious accident rates by 80%.

And while still at an experimental level, it is also being used in some natal clinics to monitor the vital signs of babies before physical symptoms of something untoward appear. This is fairly cutting-edge use of predictive analytics, and while many organisations are nudging their way towards CRM benefits, it is certainly not yet being widely adopted, says Kellett. But as HMRC illustrates, if the effort is put in, it can bring potentially enormous benefits.

Read more on Big data analytics