Big Data Part 1: How to plan to get useful (and usable!) data
Posted by Nazz Baksh
In this thought provoking two-part series by a Business World Super User, Nazz Baksh explores the topic of Big Data: what data to collect for analysis, how to satisfy current needs and how to plan for the future.
Data analysis has long been an area of discussion amongst the “numbers group” of our species – the accountants, the actuaries, etc. Only in recent years has this burgeoning skill set made it to the forefront of many organizations and gained widespread appeal. There has been much written about the explosion and availability of big data and its use to support decision-making. We all realize the merits, but the real challenge is getting started. And, how do you get started when you’re a small fish in the proverbial big pond?
But first: full disclosure. I’m no expert. I, just like you, merely have a passion for using data and employing data to support business analysis – an “armchair data user,” or a data “field-medic” – using my knowledge and available resources to develop some sort of systematic application within my organization. Very few organizations have the long-term vision to develop a proper plan for the collection, collation and organization of data. And more importantly, few organizations have the deep access to resources to do this properly. If you are lucky enough to have those resources, embrace it, in all of its splendour coupled with the headaches and seemingly endless meetings. For the rest of us, we typically have to make do with what we have.
The initial challenge is determining how to choose what you need today while planning for your future needs. That type of long-term thinking while meeting current day needs is a delicate balance, but one that should be at the forefront when creating any type of scorecard, dashboard or, to a larger extent, setting up a new financial or enterprise software solution.
The questions that ought to be asked should be: How do we know we’re on the right track? What measures accurately tell us the story of where we may be going off course? What are our key risk areas and how can we measure these risks?
These types of leading and thought-provoking questions will invariably yield metric elements and measures, some of which will already be easily retrievable and some that may need some new ways to collect.
At Peace Hills Insurance, we retired our twenty-some-odd-year-old general ledger. This was the perfect opportunity to re-evaluate how we examine our business and build something for the next 20 years (metaphorically speaking). Throughout the process, we recognized the need to collect data on our accounts in order to make good business decisions, but we also recognized that not all data is good data. Certain trade-offs needed to be made, otherwise we would be mired in seemingly perpetual stream of accounts and processes to merely collect the data, let alone analyze in a meaningful way.
Our first stop was a revamp of our chart of accounts and deciding how our business currently measures itself for performance, and how it needed to do so in the future by either establishing new accounts or developing secondary attribute features to put flavour to the nature of the activity within the account.
After analyzing and establishing clear activity-based measures, developing appropriate and automated allocation rules allowed us to appropriately code expenses to product lines, which was not done before. Previously, we struggled with a concentre method of determining profitability of a particular business line. We knew net profits about 70 to 80% of the way, but, assessing what to do with cost pools for indirect costs was a challenge. By establishing automated allocation rules, based on an activity-based methodology, we could accurately gauge the “true” profitability of the product to make better decisions without the need for additional manual work.
Understanding your purpose of collecting data
Data collection is all about providing meaningful information for decision-making. That’s at the heart of management accounting, but that needs to be tempered. You simply can’t just collect data because you can – you’d be drowning in stats with no particular purpose and, in some instances, crossing privacy lines. Your reasons for collecting data may also be varied:
- Risk assessment: What are the risks in the current environment that will prevent us from achieving our goals? What are the emerging risks and how can we continually monitor them?
- Managing growth: How can we ensure we are growing appropriately sustainably? Are we growing in areas we shouldn’t be growing in?
- External factors monitoring: What secondary measures are influencing business decisions in the marketplace such as interest rates, unemployment and labor force utilization?
- Management metrics: How can we ensure our management team is focused on initiatives that are aligned to our business plan? What metrics support this? What elements detract from it?
- Long-term strategic planning: How can we ensure our management team is focused on initiatives that are aligned to our long-term strategic plan? What metrics support this? What elements detract from it?
Assemble the right team
Once you recognize the appropriate data elements, it is critical to assemble the right colleagues to decide what information is missing and to create a plan for its proper collection and distribution. The right colleagues may be senior managers but, in some cases, may not be. Look for leaders within each functional unit of your organizations; these are the individuals with the deepest working knowledge of how systems work and what pieces of data are readily available. No experience is more deflating than to develop a set of metrics that is impossible to report on because the data does not exist or would take days or weeks to collect. Data points need to be easily collected and readily available.
Read Part 2 from Nazz Baksh of Peace Hills Insurance for Big Data Part 2: A Step-by-Step Data Collection Plan.