Successfully applying predictive analytics for maximum benefit and ROI
By David McWilliam, Director at Cortell Corporate Performance Management
Although predictive analytics is currently a hot topic, many companies are failing to realise maximum benefit and return on investment (ROI) from predictive analytics projects and implementations. One of the main reasons for this failure is that organisations purchase the technology first and then try and figure out how they can fit the solution into their business. However, for maximum benefit to be realised, this process should in fact be reversed. It is important to examine the practical applications of predictive analytics, in order to understand where it can be applied to achieve these benefits. Successfully applying predictive analytics requires a change in thinking, as well as careful planning, so that the power of this tool can be leveraged to add value in any organisation.
Predictive analytics, like business intelligence (BI) and other analytical tools, can deliver great benefit to organisations, enabling them to make more informed, fact-based decisions and improve business agility. However, businesses often fail to realise these benefits, resulting in significant technology spend that does not deliver value.
This is largely due to technology hype and once the solution is purchased, trying to shape their business and needs around what the solution is able to deliver, while scrambling to find the data they need to achieve this. This approach has several inherent problems, not least of which is that business requirements are not ‘one size fits all’ and tools don’t always deliver what the business expects or requires. In order to ensure that technology and business needs are aligned, it is first necessary to begin with the business requirements and then examine the data the business has available and how predictive analytics could benefit the organisation. Only then can you look towards a technology solution that is able to work with the available data to deliver business value.
Another issue that results in the perceived failure of predictive analytics implementations is a lack of understanding of the purpose and practical applications of the tools. While predictive analytics is highly intelligent software, it needs to be applied in the correct areas otherwise it will not deliver as expected. While BI delivers scorecards and dashboards for high-level strategic decisions, predictive analytics often solves challenges at the transactional level, which is more useful for operational business decisions. Focusing predictive analytics at the executive level has limited application, and this is an error that is often made. Predictive analytics works far more effectively in operational systems, analysing transactions and understanding customer behaviour, and this is where the true value of this software lies.
The power of predictive analytics lies in analysing transactions that are sometimes simple for people to handle in small volumes but become complex as a result of the volume and velocity with which these must be handled. People do not have the time to manually analyse these transactions, due to their sheer volumes, therefore predictive analytics can be used to help process workers to more effectively make decisions quickly with the information they have available. The software analyses both current and historical data and scores criteria to come up with suggestions on decisions quickly and simply.
For example, a claims assessor for an insurance company needs to deal with hundreds of claims a day, and must decide whether these claims are fraudulent or legitimate. Given enough time, the assessor could analyse all of the necessary information and make this decision independently, however, there simply are not enough hours in a day to enable this to be done effectively manually. Predictive analytics can use historical data from similar transactions, as well as customer data and other information, to accurately predict the likelihood of a claim being fraudulent, in near real-time. This can save insurance companies significant sums of money in paying out claims.
Call centre sales agents can be assisted by predictive analytics to identify cross-selling opportunities. Based on the historical data of the client, their demographic and psychographic profile, and the products purchased by other similar individuals, products and offerings can be tailored for maximum successful uptake. Telecoms companies can use predictive analytics to predict customer churn, by examining the behaviour of a client and comparing this to known behaviours of customers who are about the cancel their contracts, again saving money and helping to retain customers. Predictive analytics often works to solve the simplest of business problems at a purely transactional level, and understanding this can help organisations to leverage value and maximum ROI from the solution.
Identifying the business requirement up front and understanding how this fits into business processes is a recipe for ensuring successful outcomes with predictive analytics. It is then vital to find the right data, and only then choose the technology. The majority of successful applications of predictive analytics lie within the understanding of customer behaviour, in banking, retail, and financial services and so on. The technology is able to examine customer behaviour over a period of several years, and enables organisations to gain insight into how they are likely to behave in a given scenario. Understanding this, and taking the correct approach to the application and implementation of predictive analytics, is vital to leveraging maximum benefit and ROI.