Today, businesses of varying sizes, products and industries are actively applying predictive business analytics to improve customer engagement, thus resulting in increased revenues and profits. Often these businesses look to their financial planning and analysis (FP&A) group to guide, if not lead, these efforts to successful outcomes.
According to the 2017 CFO IT Survey, over 70 percent of finance executives said they plan to substantially increase the use of data analytics in the next two years, to support decision-making and improve business partnering. Fully 68 percent of respondents plan to improve their data analytics skills in the coming year. Consequently, it is essential that FP&A professionals understand what it is going to take for them to succeed.
Here are the five keys to applying predictive business analytics:
1. Framing the question or outcome: Framing the question properly will better illuminate the resources, data requirements, tools and methods that need to be selected to solve the question and provide insight to decision-making.
2. Structuring the team: It is important to bear in mind that assembling the right team is critical to success. The team structure should include key skillsets as illustrated below:
- Understand data
- Integrate and manipulate
- Know structured content and hybrid data
- Understand data quality and standards
- Know analytics
- Understand techniques and model development
- Diagnose/validate models (“Trust, but verify”)
- Perform data discovery and visualize results
- Focus on the business
- Information requirements
- Create business value from insights
- Enable management review and decision management
3. Discovering insight: Drawing insight from a piece of data involves understanding how it fits into the larger picture of an organization (i.e. context), explained Jeff Jonas, IBM’s Entity Analytics chief scientist. Business environments aren’t the only ones that require context; context is a necessity for any attempt to know more by examining data. According to Edward Deming, “Information is a means to an end. The ultimate purpose of collecting data is to provide a basis for action or a recommendation”.
During data discovery, there often is a need to refine the model results. Several techniques are particularly helpful, such as back-testing, or using historical data to test/validate known outcomes; black swans, events that happen by surprise, have a major effect, and are often inappropriately rationalized after the fact with the benefit of hindsight; grey swans, which are events that can be anticipated to a certain degree, but are considered unlikely to occur and may impact the overall market; and latency adjustments, which consist of adjusting for time shifts or unanticipated delays.
4. Applying tools and methods: FP&A project leaders should consult with their data scientist team members as to which method should be used to model the question and its possible outcomes or implied actions. For example, Netflix and Amazon use collaborative filtering, using the alternating least squares technique, and healthcare providers are applying predictive medicine using machine learning (i.e., how computers learn without being explicitly programmed). These techniques are used to support predictive analytics and more recently prescriptive analytics.
Another challenge facing project teams is the identification and collection of data. Data can be internally or externally sourced. It can also be structured, such as data values and attributes already captured in IT systems, or unstructured, such as text data extracted from social media sources. For example, Google trends or Twitter retweets can be vital sources of information and used in predictive models.
5. Sustaining organizational effectiveness: The FP&A team must recognize that the process is as much about change management as it is about analyzing data and predicting outcomes. There are numerous approaches to managing change in an organization, especially those that are large and complex.
The capability to achieve and sustain organizational effectiveness requires three important elements. First is understanding the journey. Often, companies will evolve their analytics capabilities in a progression of maturity stages.
Second is having a leadership group that can leverage the insights and the capacity to make decisions and implement needed actions. This leadership group nurtures a facts- and data-based management style that fosters a disciplined approach to performance management and decision-making and leads to actions that are aligned with business strategies and S&OP management directives.
Third is FP&A’s challenge to seek the right balance with line managers between data insight and data ownership, especially for decisions and operating results.
One final piece of takeaway advice for the FP&A professional predictive business analytics—think of analytics as a toolbox. One screwdriver of a particular size and type isn’t going to build you a chest of drawers—much less so in the hands of an inexperienced carpenter. Although some amount of predictive modeling can be done by means of preformed queries, pivot tables, summarized reports, and so on, that’s just rudimentary. The ultimate value of all that data will only become evident if the people who know how to think about it can access it easily, explore different views, test various hypotheses, and share their findings effectively.