Question first, data second: The role and practical challenge of data in decision making

Emergencies and disaster situations, remind us of the importance of good data for decision making

The volatile and uncertain environment we operate in requires rapid decision making.  The COVID pandemic, for example, is creating significant uncertainty – driving the need to monitor and react to a diverse set of changes on a daily basis.  These changes in turn drive decisions around risk, investments, supply chain co-ordination, operational planning and customer communication. More than ever leaders are seeking a robust, data-driven evidence base to make sense of their situation, and to guide decision making.

Uncertainty typically results in a cry for “more data”

When in doubt, there is often a demand to gather more data before a decision can be made. These requests for ever-more data typically take many forms: more recent, more segmented, more history, more sources, more analysis, more charts, more scenarios.

A hive of activity invested in collecting, analysing and reporting more data can give teams the illusion that great progress is being made. However, all this activity can simply be a case of misdirected effort or ‘paralysis-by-analysis’ – in reality, delaying any real progress.

Exhibit 1: Start with the question first

Opportunities can sometimes be lost while waiting for “perfect data”

Leaders and decision-makers must operate in a world of ambiguity, resource and time constraint. Waiting for data perfection is not an optimal response. Indecision and inaction can drive significant cost due to the costs caused by delay, the escalation of risks, and missed opportunities.  Leaders should acknowledge a level of uncertainty while maintaining a keen focus on reducing it where possible, in a timely manner. 

In fact, organisations that embrace uncertainty in their decision making will deliver stronger outcomes as they seek out ways to create options, plan for contingencies, and manage downside risks – without always identifying a “silver bullet” answer.

Understanding the business question first, helps clarify the few critical datapoints

In the world of big-data, data-science, the internet of things and machine learning it can be easy to be drawn to the vast supply of the latest and greatest data on offer.  However, by first taking the time to identify the demand for data, a more direct route to the highest-value information amongst the flood of available data can be identified.

A demand view of data first seeks to articulate the fundamental questions (or sets of) which will underpin business decisions. Each question can then be mapped to the relevant data, the level of detail (e.g. segmentation, time-frame and scenarios), analysis and source of data.  Many business decisions are based on the criteria that there is sufficient confidence that marginal benefits will exceed marginal costs. A go/no-go decision as such does not require data of infinitesimal error in order to make robust decisions to proceed or not.

Articulation of the business question provides a simple tool to communicate the ‘Why’ behind data

The process of identification and articulation of the business question to answer, and the supporting data to respond, creates a dialogue for consensus amongst senior executives and data analytics teams.  Being armed with a clear ‘why’ empowers data analytics teams with greater scope for innovation and efficiency in determining the best solution within project constraints.  Once the data, sources and metrics are defined for key business questions, the organisation now has a platform to repeat and extend this analysis in the future to further reduce uncertainty and provide decision makers with more sophisticated answers.

Exhibit 2: Illustrative Example: Government COVID response

Five actions can be taken to strengthen decision making in the face of uncertainty:

1. Start with the business question first
Pause the charts, reports and regression analysis until you can first articulate the business questions to be answered.

2. Communicate the question and data needs
Decision making is a team process which requires a common language between those making the decisions and those collecting and analysing the data. Business questions linked to data needs provide a simple communication tool to ensure everyone is on the same page.

3. Don’t wait for perfect
Build a data approach today, focused on the most important questions. It may not be perfect from day one, however, you will learn as you test and refine. Once validated the process can be automated and integrated into enterprise software at a later date.

4. Manage Uncertainty
Make decisions with an understanding of the level and implications of uncertainty. Data can reduce uncertainly but not eliminate it. Consider options that are robust to uncertainty: Can you stagger commitments? Can you share risk? Can you buy time?

5. Continue to build on what you have
Establish a program to continually extend and refine your data approach building upon the fundamental capabilities you have developed. Decision makers will always seek more nuanced and sophisticated analysis to further reduce uncertainty where merited.

SPP has extensive experience assisting clients to develop data strategy and decision-making frameworks.

With established consulting practices across  Public and Private Sectors, SPP has leveraged our deep project experience and subject matter expertise to support clients as they seek to integrate and embed data driven decision making in their organisation.