I recently checked out The Evolution of Decision Making – How leading organizations are adopting a data-driven culture, a new white paper from the Harvard Business Review sponsored by SAS:
People have long preached the benefits of relying on data and insights from business intelligence (BI) and analytics to help make better and timelier decisions. A reliance on data from these tools was expected to deliver better financial performance. A global survey of 646 executives, managers and professionals across all industries and geographies revealed a significant, albeit subtle, change in decision-making processes and their use of these analytics/BI tools. This report examines the results of that survey and explores how analytics is improving — and changing — the way decisions get made.
I was struck by several elements of the paper and how they demonstrated the potential value of decision modeling in building a data driven culture. I have blogged about decision-modeling before and you can find a white paper on using decision modeling in analytic projects on our website but, to summarize: A decision model takes a decision that is going to be made more than once (such as Select Marketing Offer) and describes how it should be made. It describes this decision in terms of a question (What offer should we make to this customer?) and allowed answers (Any valid marketing offer defined in the campaign management system). It identifies the information required to make the decision as well as the authorities that should be referenced to define how it can and should be made. The information required might come from data input to the decision (a Customer record say) or from other pre-cursor decisions (What product should we make an offer about?, Is the timing good for an offer?, How valuable an offer should we make?). Authorities can be policies, regulations, expertise or best practices (Marketing Know-how) as well as analytic insight (Customer Propensity to Accept an offer). All these elements are combined into a decision requirements diagram like the one at right. Each decision identified can be broken down and analyzed in turn to create a network that describes exactly how we want to make the decision. So how would such a model help you create a true, data-driven culture: Well first the paper identified several aspects to becoming more analytic, more data-driven, among decision makers. Decision modeling it seems to me could help in each of them:
- Enhancing skills to include analytics in normal working One of the challenges in including analytics in normal working is that most companies have never defined what “normal working” means for the decisions they make every day. A decision model would do so.
- Balancing data with instincts A decision model does this explicitly. Instead of relying on people to remember to use the right mix of authorities for a decision, a decision model would lay this out precisely – at every level you can see what instincts/best practices/expertise should be combined with what kind of analytics (and what regulations or policies must be followed).
- Building relationships with analytic professionals I am a firm believer that analytic success requires business, IT and analytic professionals to work together. The great thing about these diagrams is that they can be built by and discussed by all three groups. A shared understanding of decision-making is the result.
- Developing best practices It’s hard to develop best practices in decision-making without a way to document them. Decision models document decision-making best practice.
See the recording of our 30′ webinar – 4 Ways Decision Modeling Creates a Data-Driven Culture?
It should also be noted that continual refinement and greater use of analytics in real-time decision-making both came up also. Continual refinement is predicated on knowing how you do something now (so you can improve it) and the role of a decision model in this is clear. Similarly using analytics in real-time decision making requires an ability to specify the requirements (so IT can built the real-time system) and decision models are ideal for this. The paper identifies several things that generate enthusiasm in companies for a change to more data-driven decision-making. These too show the potential value of decision modeling. The pressure of compressed decision-making time frames was cited repeatedly. This need to make decisions more quickly is likely to be focused on those decisions that are made most often. It is these repeatable decisions where a decision model pays off in terms of improved automation (the model lets you see which bits of the decision can be automated) and repeatability. At the same time respondents noted that many decisions lack transparency. This, of course, is a no-brainer. We don’t know how we make decisions because we don’t document how we make decisions. A decision model gives you this design-transparency. Finally the paper identifies a number of traits for companies that are analytic leaders and the top 3 can all be supported by decision modeling:
- Executives mandate use of analytics and an analytic decision-making process A company that uses decision modeling can share a best practice-based approach to making decisions, sharing a common vision of how each decision will be made and the role of analytics in that decision.
- Use the right metrics A decision model can be (and should be) mapped explicitly to your key metrics, allowing a company to see which decisions impact which metrics. This helps ensure you are using the right metrics and means that using the right metrics can directly improve decision-making.
- Decision-making transparency If you don’t have a way to describe decision-making, and a way that works for everyone (business, IT and analytic experts), then how can you achieve transparency? Decision modeling gives you the transparency you need.
To see what I mean, why not view the recording of the webinar we did – 4 Ways Decision Modeling Creates a Data-Driven Culture? Just 30 minutes to show how decision modeling helps create a data-driven culture. You can also download our white paper on decision modeling for analytic projects here and sign up to try out our cloud-based decision modeling software DecisionsFirst Modeler?
You can download the HBR white paper here (registration required).
Comments on this entry are closed.
You’re absolutely correct. Making complex decisions without a model is like trying to navigate a foreign country without a map. You end up in places you really don’t want to be. One of the biggest mistakes a financial institution (FI) can make is thinking that a piece of technology will automatically make their decisions for them. Just like installing a firewall without configuration will do little to protect your data, technology is only as good as the decision model. We saw a few years back the impact of automating bad decisions. Whether an FI is looking to implement multichannel cross-sell or instant credit decisioning, there must be a solid, well-planned model to back it up.