Analytical correlations and prospects are only meaningful on operational data. Applied to purely financial figures, the accounting reflection of these operations, they bring little to no grist to the mill.
Within its reporting environment a CFO today has dashboards with many KPIs, ideally periodically redefined, data visuals with graphics, trends and pie charts, summarized and detailed PowerPoint and Excel spreadsheets.
This dashboard is automatically or semi-automatically fed with internal business data, whether or not designed with the help of hired IT or business consultants. Progressive CFOs also add some valuable external data to it.
What is the next step ? A projection into the future.
The questions a CFO should ask herself are: What is my objective? What will I do with it? Up to where am I willing to go? Is real-time reporting really needed for my own requirements? Her ultimate goal must simply be to provide added value by better forecasts.
It goes without saying that the activities of the company and its market play an important role in the way to find those answers. A production environment with volatile material flows is quite different from a services’ company only producing paper forms.
1/ For a meaningful future projection, the CFO must first search for historical trends, links and explanatory business drivers. Everything usually has trickle down effects. E.g. a significant increase or decrease in the price of raw material speeds up or slows down the frequency of deliveries by the suppliers. In addition, it might also trigger exercising specific hedging contracts.
Scenario analysis, what-if's, A / B testing, customer segmentation, churn and cohort analysis are all examples of analyzing historical operational data. Eventually, they will obviously each separately have an impact on the financial figures.
Let us not forget a common and recurring topic here: the preventive maintenance of specific assets’ categories, together with their direct impact on related business operations and financial variables.
2/ Data enrichment, incorporating the right external information in the study to avoid a tunnel vision, is the next required step in the search for reliable forecasts. This would be from market penetration rates to economic growth expectations, past and future demographic trends and so more. Do note that sometimes the expected resulting impact might be very vague, unclear or simply wrong sometimes (think Brexit, the US presidential elections).
The CFO must be fully aware on the limits of the "expected occurrence power" of her calculated projections, to avoid a potential risky and in hindsight wrong strategy.
3/ Both elements above are instrumental in evaluating the predictive power of the available data and deductions. It is the job of the CFO now to understand the correlations, prove them and then use them as guidance, in close cooperation with the statistical or mathematical profiles.
‘Analyzing to simply analyze’ offers little added value. Financial figures and balances in general ledger accounts are only a reflection of the underlying business operations.
Predictive Analytics should, whether or not combined with big data and AI, be added asap to the risk management agenda of any CFO.
The purpose is to apply it where it is relevant and potentially has the most impact. Today, and probably for 2017 too, that still seems to be on the operational figures, straight from the "factory floor", and not on the subsequent numerical financial translation.