By The Open Group
Most enterprises today are in the process of jumping onto the Big Data bandwagon. The promise of Big Data, as we’re told, is that if every company collects as much data as they can—about everything from their customers to sales transactions to their social media feeds—executives will have “the data they need” to make important decisions that can make or break the company. Not collecting and using your data, as the conventional wisdom has it, can have deadly consequences for any business.
As is often the case with industry trends, the hype around Big Data contains both a fair amount of truth and a fair amount of fuzz. The problem is that within most organizations, that conventional wisdom about the power of data for decision-making is usually just the tip of the iceberg when it comes to how and why organizational decisions are made.
According to Penelope Gordon, a consultant for 1Plug Corporation who was recently a Cloud Strategist at Verizon and was formerly a Service Product Strategist at IBM, that’s why big “D” (Data) needs to be put back into the context of enterprise decision-making. Gordon, who spoke at The Open Group Boston 2014, in the session titled “Putting the D Back in Decision” with Jean-Francois Barsoum of IBM, argues that a focus on collecting a lot of data has the potential to get in the way of making quality decisions. This is, in part, due to the overabundance of data that’s being collected under the assumption that you never know where there’s gold to be mined in your data, so if you don’t have all of it at hand, you may just miss something.
Gordon says that assuming the data will make decisions obvious also ignores the fact that ultimately decisions are made by people—and people usually make decisions based on their own biases. According to Gordon, there are three types of natural decision making styles—heart, head and gut styles—corresponding to different personality types, she said; the greater the amount of data the more likely that it will not balance the natural decision-making style.
Head types, Gordon says, naturally make decisions based on quantitative evidence. But what often happens is that head types often put off making a decision until more data can be collected, wanting more and more data so that they can make the best decision based on the facts. She cites former President Bill Clinton as a classic example of this type. During his presidency, he was famous for putting off decision-making in favor of gathering more and more data before making the decision, she says. Relying solely on quantitative data also can mean you may miss out on other important factors in making optimal decisions based on either heart (qualitative) or instinct. Conversely, a gut-type presented with too much data will likely just end up ignoring data and acting on instinct, much like former President George W. Bush, Gordon says.
Gordon believes part of the reason that data and decisions are more disconnected than one might think is because IT and Marketing departments have become overly enamored with what technology can offer. These data providers need to step back and first examine the decision objectives as well as the governance behind those decisions. Without understanding the organization’s decision-making processes and the dynamics of the decision-makers, it can be difficult to make optimal and effective strategic recommendations, she says, because you don’t have the full picture of what the stakeholders will or will not accept in terms of a recommendation, data or no data.
Ideally, Gordon says, you want to get to a point where you can get to the best decision outcome possible by first figuring out the personal and organizational dynamics driving decisions within the organization, shifting the focus from the data to the decision for which the data is an input.
“…what you’re trying to do is get the optimal outcome, so your focus needs to be on the outcome, so when you’re collecting the data and assessing the data, you also need to be thinking about ‘how am I going to present this data in a way that it is going to be impactful in improving the decision outcomes?’ And that’s where the governance comes into play,” she said.
Governance is of particular importance now, Gordon says, because decisions are increasingly being made by individual departments, such as when departments buy their own cloud-enabled services, such as sales force automation. In that case, an organization needs to have a roadmap in place with compensation to incent decision-makers to adhere to that roadmap and decision criteria for buying decisions, she said.
Gordon recommends that companies put in place 3-5 top criteria for each decision that needs to be made so that you can ensure that the decision objectives are met. This distillation of the metrics gives decision-makers a more comprehensible picture of their data so that their decisions don’t become either too subjective or disconnected from the data. Lower levels of metrics can be used underneath each of those top-level criteria to facilitate a more nuanced valuation. For example, if an organization needing to find good partner candidates scored and ranked (preferably in tiers) potential partners using decision criteria based on the characteristics of the most attractive partner, rather than just assuming that companies with the best reputation or biggest brands will be the best, then they will expeditiously identify the optimal partner candidates.
One of the reasons that companies have gotten so concerned with collecting and storing data rather than just making better decisions, Gordon believes, is that business decisions have become inherently more risky. The required size of investment is increasing in tandem with an increase in the time to return; time to return is a key determinant of risk. Data helps people feel like they are making competent decisions but in reality does little to reduce risk.
“If you’ve got lots of data, then the thinking is, ‘well, I did the best that I could because I got all of this data.’ People are worried that they might miss something,“ she said. “But that’s where I’m trying to come around and say, ‘yeah, but going and collecting more data, if you’ve got somebody like President Clinton, you’re just feeding into their tendency to put off making decisions. If you’ve got somebody like President Bush and you’re feeding into their tendency to ignore it, then there may be some really good information, good recommendations they’re ignoring.”
Gordon also says that having all the data possible to work with isn’t usually necessary—generally a representative sample will do. For example, she says the U.S Census Bureau takes the approach where it tries to count every citizen; consequently certain populations are chronically undercounted and inaccuracies pass undetected. The Canadian census, on the other hand, uses representative samples and thus tends to be much more accurate—and much less expensive to conduct. Organizations should also think about how they can find representative or “proxy” data in cases where collecting data that directly addresses a top-level decision criteria isn’t really practical.
To begin shifting the focus from collecting data inputs to improving decision outcomes, Gordon recommends clearly stating the decision objectives for each major decision and then identifying and defining the 3-5 criteria that are most important for achieving the decision objectives. She also recommends ensuring that there is sufficient governance and a process in place for making decisions including mechanisms for measuring the performance of the decision-making process and the outcomes resulting from the execution of that process. In addition, companies need to consider whether their decisions are made in a centralized or decentralized manner and then adapt decision governance accordingly.
One way that Enterprise Architects can help to encourage better decision-making within the organizations in which they work is to help in developing that governance rather than just providing data or data architectures, Gordon says. They should help stakeholders identify and define the important decision criteria, determine when full population rather than representative sampling is justified, recognize better methods for data analysis, and form decision recommendations based on that analysis. By gauging the appropriate blend of quantitative and qualitative data for a particular decision maker, an Architect can moderate gut types’ reliance on instinct and stimulate head and heart types’ intuition – thereby producing an optimally balanced decision. Architects should help lead and facilitate execution of the decision process, as well as help determine how data is presented within organizations in order to support the recommendations with the highest potential for meeting the decision objectives.
Join the conversation – #ogchat
Penelope Gordon recently led the expansion of the channel and service packaging strategies for Verizon’s cloud network products. Previously she was an IBM Strategist and Product Manager bringing emerging technologies such as predictive analytics to market. She helped to develop one of the world’s first public clouds.
Strongly agree Penelope. Of course that’s what enterprise architects should be doing anyway – not developing data architectures. Anyway, it’s great to see this in print.
Absolutely. You make an excellent point and it makes an interesting read. Mostly the challenge is turning data into useful and usable information and providing that to the people that can use it for the benefit of the business.
Comments are closed.