The Purpose of Analytics is not Reporting or Monitoring but Deciding

Some years ago we did some research on the landscape of analytics capabilities. The research looked at the increasingly broad portfolio of analytic capabilities available to enterprises – everything from traditional Business Intelligence (BI) capabilities like reporting and ad-hoc queries to modern visualization and data discovery capabilities as well as advanced (predictive) analytics. One of the motivators for the research was that these capabilities are often presented as part of a maturity curve, where enterprise customers are expected to keep moving “up” the curve to get more value. We found, in fact, that enterprises adopt these capabilities in a variety of sequences and ultimately need a mix of them. The real questions are what situations need which capabilities and who is the target user for these capabilities.

While there seem to be as many reasons for adopting analytic capabilities as there are organizations adopting analytics, the reality is that three key business needs are driving analytic adoption – reporting, monitoring and deciding:

  • Reporting
    Most organizations need to report on some aspect of their operations. To meet a reporting need, an organization must present some or all of the data it has gathered in a report to some internal or external body. For instance, the Dodd-Frank act requires that all financial trades meeting certain criteria are reported to the federal government, and most organizations with hourly workers need to provide employee hours to HR. These reports are generated for compliance or policy reasons to someone who generally is authorized to request data at a specific level of detail or timeliness. The organization that produces the report is only required to provide that data in report form. Reports are often constrained by circumstances and delivery style.
  • Monitoring
    Most organizations want to monitor their behavior or performance. Generally, an organization identifies metrics or key performance indicators (KPIs) and each department receives the tools necessary to monitor their metrics. For instance, an organization might have a group of managers and executives that track total sales, customer retention rates, and average customer profitability. Monitoring can rely on reports issued over time, but it generally uses graphical or visual dashboards that clearly illustrate how the metrics change from baseline values over time.
  • Deciding
    Organizations increasingly see value in making data-driven or analytic decisions. This need can be explicit, identified at the very beginning as the rationale for an analytic capability. It can also be implicit, as organizations that think they need reporting or monitoring realize that it is acting on the reports and monitoring dashboards that is critical to success. For instance, the organization might need to decide who to schedule for which shift, which customers should get a pre-emptive retention offer or what discount to offer on a specific order.

By the way, reporting is not the only use case for reports – a report may be used to monitor performance and acting on a report is a decision-making scenario, not a reporting one. Similarly, taking action on the basis of what is monitored is also a decision-making scenario.

Historically, the focus for most organizations has been reporting. Over time this shifted to a balance of reporting and monitoring. In the years since the study, however, the focus of analytic investment has shifted hard to decision-making. Indeed, a poll taken at the time shows that those working in analytics anticipated this shift. Three quarters of respondents were focused at the time on reporting or monitoring but fast-forward 12-24 months, and almost eighty percent of respondents expected to be focused on decision-making instead.

And it is decision-making – not reporting or monitoring – where analytic capabilities and software can add value. If you are spending money on analytics for reporting or monitoring then you are probably wasting most of it.

  • Reporting is well-defined and largely straightforward until there is a need to make decisions based on the content being displayed when the complexity and value comes from the decisions being made. Analytic software may make it faster and cheaper to produce a report but this shows a limited ROI for everyone outside IT.
  • Monitoring is also pretty straightforward unless the person doing the monitoring needs to make a decision based on what they see. At that point, the complexity comes from the decision(s) they have to make, rather than the monitoring itself. Analytic software often looks like it has an ROI in monitoring but the use of more sophisticated analytics like seasonality, forecasting and predictions in monitoring are really helping with decisions made based on the monitoring, not the monitoring itself.

What matters is decision-making. Using analytic software to improve the quality of decision-making, making more accurate and profitable decisions, is where the ROI truly lies.

And this is where organizations get into trouble because they think they can improve their decision-making – make it more data-driven – without being clear about the decision or decisions being improved. They start by looking at their data. They seem to believe that if they pile up enough data, clean it enough, integrate it and buy enough tools for their “citizens” to use, that they will see an ROI. They won’t. It doesn’t matter how pretty the visualizations are, how fast the reporting is, or how clever the algorithm gets. If you don’t understand the decision-making you are trying to improve then it’s all just a way to spend money on new toys.

If you want to get value from analytics, you need to focus on decision-making. This means focusing on specific decisions that you can name, describe, model and understand. If you don’t know what decision-making you want to do better, and cannot describe what “better” means, then any money you spend on analytics is likely to be wasted. And the more sophisticated the analytics get, the more likely this becomes. Machine Learning and AI models require more, not less, understanding of the decision being improved.

As we like to say here at Decision Management Solutions, put DecisionsFirst™. Not data, not reports, not dashboards. Decisions.

 

You can get the full Analytics Capability Landscape report and an accompanying infographic here. It’s got lots of good content on the different kinds of decisions you might want to use analytics for and some suggestions for aligning capability with decisions and roles.

 

One Comment

  1. Christopher Davis says:

    Thank you. This is my new manifesto. I have been telling my team and our stakeholders that we need to start thinking about data from the perspective of what decision is going to be made and then work backwards. This article articulates that perspective wonderfully.