PAI: How do you see the predictive analytics market develop over the next couple of years?
PH: We’re entering the era of the predictive enterprise. It’s a broad concept and one that’s been talked about a lot in recent months, but I believe there are going to be some interesting changes in the corporate space. Businesses are now looking at very real time requirements. On the one hand, there is the possibility for a complexity and volume of transactions not seen before. These are the kinds of requests that need to be handled by dedicated event processing engines. On the other, there will be instances of lower throughput that embedded capabilities will match, such as predictive claims processing in the insurance sector, and predictive campaigns for marketers.

PAI: What new trends and developments will become increasingly important in the PA market?
PH: While real-time analytics is still quite new, even for capital markets, it will take some time to bed in. The implications of this kind of process are still being felt in areas like the data warehouse (such as turning them into active data warehouses). In order for true predictive analytics to take off, information needs to be built into the platform that is being used, whether event-based or more traditional. Bespoke processes are being applied to regular queries, but much of the technology is still buried in the back end. Further ahead, predictive analytics will be much more about front-end analysis. It won’t just be historical analysis from data in the warehouse: it will be real-time predictions, developed by front-end applications and delivered immediately to users.

PAI: How does PA relate to BI in your view – when should one look at what type of product?
PH: Almost everything that happens in an organisation is related to some sort of pattern. If it’s a new customer order, a spike in sales, a member of staff’s resignation or even a drop in response times, you need to know whether that represents a threat or an opportunity. That’s where predictive analytics works, because it provides a context within which these events occur: for example, allowing you to analyse and compare today’s sales with the same time last week or last month or last year, and then projecting forward to what is most likely to happen next week or next month, or even tomorrow. Business intelligence tools on the other hand, are good at accessing information about specific incidents and analysing it from a historic perspective. PA and BI can be used to reinforce one another but they are essentially different.

PAI: Do you think the role of PA within organisations is changing – and how?
PH: Predictive analytics is starting to become far more important to businesses. Organisations are beginning to ask “Am I agile enough?” Anticipating future events is key to commercial survival. People are realising that the better you can predict the future, the better placed you are competitively.

PAI: What do you view as the most important business issues to solve with PA?
PH: The business issue is definitely dependent on the industry, but there are some constants. For telecommunications companies, for example, predictive capability is critical for understanding customer behaviour, either so that you can avert churn or so that you can up/cross-sell. In capital markets, financial predictions are a classic example of where the technology is really embedded. Also, retail is an industry where trends, both consumer and manufacturing, have a significant impact. This is therefore an area where it is critical to have a good level of understanding and analysis.

There are also some less expected areas where predictive analytics are valuable – I know of an Italian football club using it to predict player performance and the likelihood of injuries. The application of the technology really depends on the context in which it’s being used.

PAI: What would you recommend to companies aiming to purchase and implement PA in their organisation?
PH: The first step would be to put a solid return on investment model in place. Build a business case, and pick the areas of the organisation that are going to demonstrate a significant return. That will then ensure you have the proof of ROI that you can then use as a model to demonstrate to the rest of the company.

Use your knowledge of the business to see where predictive analytics should go first, as a successful project in one area is far more valuable than a company-wide project that was too ambitious from the outset – a valuable predictive analytics solution depends on the intelligent application of technology to the right area.

PAI: What do you think of the importance of unstructured data within organisations?
PH: Unstructured data is much more important than most organisations think. As businesses have grown in their use of technology, my experience suggests that companies generally have far more data than they think they do. Historically, that data hasn’t been exploited. If you could look at that data, spot patterns in it that tell you a lot about your business, and then use those patterns to make predictions, that is a valuable business tool. The information that is found is often a surprise – some companies find that they’ve been investing in the wrong areas, such as building customer acquisition when retention should be the focus.

PAI: What do you think of the importance of customer feedback data? How should organisations make more use of it?
PH: If you are getting extensive feedback from customers, that’s a great start. But analysing that data correctly is critical to getting the most return. You need to make sure the data is valuable, and coming from the right groups. Then you should ensure you are making decisions based on the right business objectives. A hard question to face is “Are these customers valuable?”, but analysis should demonstrate what you know and what you should do. The answer often lies in the deeper calculations that predictive analytics can offer.

About Philip Howard

Philip started in the IT industry in 1973 and has worked as a systems analyst and programmer, as well as in sales, marketing and product management, for a variety of companies including GEC Marconi, GPT, Philips Data Systems, Raytheon and NCR.

Working with Bloor Research since 1992, Philip is now Research Director. While maintaining an overview of the data and content market, Philip himself specialises in databases, data management, data integration, data quality, data federation, master data management, data governance and data warehousing. He also has an interest in event stream/complex event processing.

He contributes to various magazines and has written a number of reports published by companies such as CMI and The Financial Times, in addition to his work with Bloor Research.