Do you understand what data analytics really is, or just that it’s a hot topic you’d like to know more about?
Getting to grips with business analytics (to give it its other name) should be near the top of the “to do” list for most finance professionals. But a lot of accountants don’t actually know what analytics is and how it fits into the accounting and finance function.
So if that is you, then this first instalment of a 5 part series on Business Analytics is a perfect way to get started.
What is data analytics?
Simply put, analytics is a structured approach to data-driven problem solving. In the context of our work, it’s using data to build models that increase your insight, leading to better decision making, and added value to individuals, companies and organisations.
The concept was first introduced to the world of business in the noughties, through the pioneering work of Thomas Davenport and Jeanne Harris. Their Harvard Business School book Competing on Analytics brought a largely academic subject to the world of business.
Analytics has become a competitive weapon due to increased processing power, huge reductions in IT storage costs, and the availability of usable data.
The shift has been so dramatic that some analysts have called data the new oil, pointing to huge performance gains and investment rewards if companies can extract it, transform it and use it to answer their most compelling questions.
Netflix uses data analytics rather than funding pilots in order to make decisions about new shows. It takes an analysis of which actors and directors customers passionately like and combines it with detailed data about scenes viewers watch compulsively or skip. This has enabled an 80% success rate for new shows and led to a string of major hits such as House of Cards and Stranger Things.
The core of what we do in accounting and finance is reporting: the structured presentation of transactional data so it is easy to understand.
Then we have business intelligence (BI), the grown-up version of reporting. BI has created a stronger link between data and decision making by providing more information and clarity to the consumer.
BI includes elements of querying data, visualisation of data in dashboards and scorecards, and even new data from simple mathematical models (ie: driver-based budgets- see glossary).
Generally BI is limited to using historical data, basic maths and internal sources of data, that falls short of predictions and recommendation that could significantly impact decision making.
If BI is grown-up reporting, then analytics could be considered to be BI’s older sibling.
Analytics uses logic, reasoning, critical thinking and complex quantitative methods to create accurate predictions and recommendations. It goes beyond the presentation and query of internal data. It’s a focus on problem-solving and answering specific business questions, using new methods and relevant available data.
BDO uses data analytics to spot risk and fraud in the audit process, and as a way to improve business performance.
Analytics can help identify ways to save money and increase operational efficiencies, such as revising payment terms, reducing duplicate payments, or aggregating payments to the same vendor.
Different types of data analytics
There are four generally accepted types of analytics. Despite the new terminology, accounting and finance already deliver the two less complex, possibly less valuable, types.
1. Descriptive analytics: what happened?
Descriptive analytics answers the question: “what happened?” Most financial reporting teams would try to produce this on a frequent basis.
One example in retail could be informing a business unit manager of what their daily or weekly sales were, or what their top 10 products were, based on an analysis of average weekly revenue for each product group.
Descriptive analytics process transactional data from multiple data sources, mostly internal, to give potential valuable insights into past performance. But as previously stated, this information is only a status update; an indication that something is more or less than expected, without any further explanation or insight.
2. Diagnostic analytics: why did it happen?
In most financial reporting teams, this is your reason for being. The time you have available after producing period end reporting and analysis packs will be spent drilling down into data, finding exceptions or differences in trends and adding commentary to answer why whatever happened has happened.
Following on from the previous retail example, you could drill down to product gross profit in each category to find out why they missed their net profit forecast, and review the forecast drivers to see why the outcome was different.
In analytics terms, neither diagnostic or descriptive analytics are considered to be hugely valuable to the problem solving process, so companies are looking for more.
3. Predictive analytics: what happens next?
Predictive analytics is used to create accurate estimates to help paint a picture of what is likely to happen.
It uses the business data normally found in descriptive analytics to detect patterns and exceptions to predict future trends, which makes it a valuable tool for forecasting. Predictive analytics is powered by machine learning.
Despite numerous advantages that predictive analytics brings, it is essential to understand that forecasting is just an estimate. Its accuracy depends highly on data quality and the stability of the environment, so it requires a careful treatment and continuous optimization.
By creating accurate analytics information, consumers can be more proactive in their decision making, to solve potential problems as the information is received.
4. Prescriptive analytics: what to do about it
This is the most valuable, taking predictions and making recommendations for the actions to take based on that prediction.
This requires both historical internal data and external information, due to the nature of statistical algorithms. Prescriptive analytics uses sophisticated tools and technologies such as machine learning, business rules and algorithms. This makes it tricky to implement and manage. That is why, before deciding to adopt prescriptive analytics, a company should compare the required efforts to its expected value.
Where do data visualisations come in?
Data visualisation is relevant to all analytics. There is no point creating new data and insight if the one who will consume it cannot understand what is being presented.
Graphs and charts are visual methods of presenting data, but visualisation goes further.
Consider the way humans process information: how best to present complex data as fast as possible, rather than too much information buried in volumes of reports and spreadsheets.
Visualisation is designed to ensure quick, easy understanding of analytics, communicated in a standardised way. Identify areas that need attention or improvement and position them so people are drawn to them first, creating a flow of visual information based on the message the analytics provides. Consider the use of formats and colours.
As you can see, there are multiple concepts and methods to consider as you mature your analytics capability, but where does that leave our core tools – MS Excel and financial modelling?
Excel has basic data visualisation, but it also has a number of mathematical models that can be used to create predictions and new data. Add-ons like Power Pivot and BI are proving very popular and helpful when starting with analytics.
That said, with the advent of big data, an ever-growing number of data sources, and the computing power of specific analytics tools, Excel can only go so far.
Analytics is a natural next step for finance and accounting’s core capability. It can enable better decision making and finance business partnering within organisations. No matter where you are on your data journey, you are probably already doing some aspects of analytics, and through a shift to predictions and recommendations enabled by new talent and tools, you can become a data-driven problem-solver in your business.
Data Visualisation is the term used to describe the way data is presented in an eye-catching way as fast as possible, as opposed to stark figures or noisy charts. It can help to engage and consumer of information and highlight patterns, trends and correlations that might not be as clear if presented poorly or in simple data.
Driver-based planning focuses on the factors that are most critical to powering the success of a business. Mathematical models are built to allow different scenarios to be simulated based on changes to these business drivers.
Machine Learning is the scientific study of algorithms and statistical models that computer systems use to effectively perform a specific task without using explicit instructions, relying on patterns and inference instead. It is seen as a subset of artificial intelligence. Eg: Look at historical payment trends to predict which customer will pay and when.
Methods and Tools; Some examples of Analytics programming languages include R and Python
Some examples of the data visualisation tools include SiSense, Tableau, Qlikview, Domo.
David Nunn is Content Manager at AAT.