From Errand to Delivery

In Chapter 1, we established that the goal of analytics is to reduce uncertainty. In Chapter 2, we discussed the impetus for working with data that comes from the strategic level of the organization and how that creates an "Analytics Errand"—the disciplined process of translating a vague strategic desire into a concrete data requirement.

Now, we arrive at the destination of that errand. The analyst has queried the database, the data scientist has trained the model, and the engineer has built the pipeline. But the errand is not complete until something is delivered back to the business.

This chapter defines what that "something" is. It is the tangible artifact—the Output—that the organization can hold, see, and use. However, as a manager, you must navigate a critical distinction that often confuses leadership teams: the difference between the Output you control and the Outcome you desire.

The Critical Distinction: Outputs vs. Outcomes

In the rush to become "data-driven," organizations often conflate the deliverable with the result. It is vital to separate them.

• Outputs are the tangible deliverables of the analytics activity. They are direct, measurable, and entirely within the control of the analytics team. A dashboard, a predictive model, a slide deck, or a cleaned dataset are all outputs. We can guarantee their delivery, their accuracy, and their timeliness.
• Outcomes are the distal business results generated when those outputs are used effectively. Increased market share, higher profit margins, or reduced churn are outcomes.

The Manager’s Reality Check:

You can build the perfect churn prediction model (Output). It might identify at-risk customers with 95% accuracy. However, if the marketing team uses that model to send a terrible offer, or if a competitor slashes prices by 50% the same week, the churn rate may not drop. In this scenario, the analytics team succeeded in their output, but the business failed in the outcome.

Outcomes are influenced by externalities—competitor behavior, economic downturns, and execution errors—that analytics cannot control. Therefore, while we aim for outcomes, we manage and measure Outputs.

The Menu of Analytics Outputs

When you commission an analytics project, you should expect to receive one (or more) of the following six distinct outputs.

1. Reports and Dashboards

This is the most common form of analytics output, often associated with "Descriptive Analytics." Dashboards serve as the instrument panel for the organization, aggregating vast amounts of transactional data into digestible visuals.

• The Lag Trap: Traditionally, reports focus on "Lag Information"—metrics like last month's sales or last quarter's churn. While necessary for compliance and monitoring, modern dashboards must strive to include "Lead Information"—indicators that predict future performance, such as pipeline velocity or customer sentiment trends.
• The Modern Shift: We are moving away from static PDFs emailed on Mondays. The contemporary standard is interactive, self-service dashboards (using tools like Tableau, PowerBI, or Looker) where the "Prosumer" manager can drill down into the data without requesting a new report from IT.

2. Ad-Hoc Analysis and Strategic Recommendations

Not every business problem requires a permanent dashboard. Sometimes, a manager faces a specific, one-time strategic decision: “Should we acquire Competitor X?” or “Why did sales plummet in the Midwest last Tuesday?”

The output here is a Strategic Recommendation. It is often delivered as a presentation or a memo, not a software tool.