About This Guide
This guide is intended for Decision Lens Super Users and Analysts. It aims to enable these users to quickly learn how to apply cost data and related metrics in decision portfolios to deliver more insights. Exactly how cost data is applied depends on the type of use case and key stakeholder objectives. This guide presents several scenarios and offers best practices for addressing each as quickly and easily as possible.
The Role of the Decision Analyst
When one hears the term “analytics”, it typically conjures an image of a handful of brainy data scientists pouring over statistical programming to weave together disparate massive data sets to find correlations between consumer purchase behaviors, or correlations between securities prices and other market phenomena. In general, the popular view of analytics is that it is all about discovering repeatable patterns in data that is far too extensive or complex for the human brain to comprehend without the aid of computing.
These characterizations are certainly valid, but are not indicative of our brand of analytics at Decision Lens. Decision Lens users also seek insights in data, but Decision Lens excels at taking a large array of potential choices that overwhelm unaided human cognitive capabilities, and narrowing them down, through a systematic process of prioritization, to a relative few choices where human judgment can be effectively applied.
Some might describe it as “boiling the ocean" or “peeling the onion.” Regardless, the ultimate objective of analytics within Decision Lens is to simplify complexity by restructuring and simplifying myriad choices available to decision makers.
As a decision analyst, you will be expected to take ownership of preparing and inputting data to help inform decision models. But you will also have a significant role in "telling the story" to guide stakeholders to make the best decision possible, and glean new insights from the data. The scenarios and visualizations you produce and present will have a significant impact on framing the decision and guiding the team.
- Do your planning homework. Determine the type of decision (e.g. simple prioritization and selection, budget allocation, resource scheduling) first, and plot your analytics approach around it. Is there reliable resource cost data? Are there supporting performance metrics that can be applied? What sample reports from past decisions are available that speak to potential criteria, and data to inform them?
- Get the data as early as you can. Conversations about required and available data should start with the decision kick-off. There is typically some lead time needed to get data pulled and prepped, so build this into your decision plan.
- Give Stakeholders guidance. Decision Lens can be unfamiliar territory for some. At its heart, Decision Lens is simple a method of benefit - cost analysis where benefits are expressed in relative terms (using criteria and ratings). Stakeholders may not readily see this, and thus analysts should take care to explain how criteria, scales, ratings and metrics work together to transform stakeholder inputs into decision insights. Teach as you go, explaining the method behind priority weights, ratings votes, metrics, Value ROI, and optimization.
- Keep it simple. Stakeholders are turned off by extensive data preparation, lots of scenarios, “noisy” visualizations, and lots of moving parts in optimization. Try to avoid any analytics that require spreadsheet-based manipulation and other external aids. This complicates the process and creates additional work that is typically not necessary.
- Don't let "perfect" overrule "good enough." Most decisions are inherently a bit "messy." Data is rarely perfect. There is subjectivity involved. Resource cost data is derived from forecasts and estimates, and almost always can be called into question. Stakeholders must be willing to live with some uncertainty as they make their decision. To the fullest extent possible, ensure that decision data is reliable and relevant, but do not let a goal to create the perfect model delay the process.