Over the coming weeks, I’ll explore each of the six pillars of the Decision Science Pre-Analysis Framework, showing how to optimize every phase to overcome the last-mile problem and turn insights into actionable outcomes that drive real business impact.

Why the “Insights Industrial Complex” is Failing Decision-Makers

We are producing more insights than ever—but are we making better decisions?

Dashboards overflow, reports pile up, and analytics teams churn out endless metrics. On the surface, this looks like progress—more data-driven, more informed, more sophisticated.

But scratch beneath the surface, and a critical issue emerges: many organizations aren’t using insights to drive action; they’re using them to showcase activity.

Instead of enabling better decisions, insights have become a performance metric of their own—proof that analytics teams are busy, dashboards are being maintained, and reportsh are being generated. But being busy isn’t the same as being effective.

The Problem: Reporting Is Not Progress

• Dashboards are descriptive, not directive. Without context or prioritization, they leave decision-makers guessing at what matters most.
• An overflow of insights leaves leaders overwhelmed. More data doesn’t simplify decisions—it complicates them without clear guidance.
• Analytics teams are measured by output rather than outcomes. Success is often defined by the volume of insights created, rather than whether those insights lead to meaningful change.

This is the “insights industrial complex” at work—a system of tools, practices, and mindsets that prioritizes producing insights over applying them. And at the heart of this issue lies the last-mile problem—the critical gap between generating insights and turning them into real-world decisions.

The Last-Mile Problem: Insights That Don’t Drive Decisions

The last-mile problem is a concept borrowed from logistics: the hardest part of delivering a package isn’t getting it across the country; it’s getting it to your doorstep.

In analytics, the last-mile problem is similar: the hardest part isn’t generating insights; it’s getting them applied to real-world decisions.

Think about how often this happens:

• A deep-dive analysis uncovers a key finding, but it never gets acted on.
• A quarterly business review presents tons of metrics, but the business keeps running on gut instinct.
• A new customer segmentation is built, but marketing never uses it in their campaigns.
• A predictive model ranks high-value prospects, but sales teams ignore it in favour of intuition.
• An A/B test shows a clear winning strategy, but the team hesitates to scale it because it’s s just one test.

The insights are accurate, but behavioural inertia prevents adoption. They never cross the finish line.

Why?

• Too many insights, not enough strategic framing. Decision-makers lack real clarity on the potential impact of their decision—or the consequences of inaction.
• Insights are produced to justify past decisions, not shape future ones. Instead of fueling innovation, analytics often becomes a tool for post-rationalizing existing strategies.
• No accountability for applying insights. It’s not enough to generate insights; businesses need a culture that demands action.

If analytics is truly about making better decisions, we must solve the last-mile problem before it starts.

Solving the Last Mile Problem Before It Starts

The biggest predictor of whether stakeholders will act on an insight is whether action was expected from the beginning.

Getting past the last mile of analytics—where insights actually drive action—is determined long before the data is even analyzed. The foundation for a successful, actionable analysis is laid at the very beginning, in how the problem is framed, how stakeholders are engaged, and how decision-making is structured.

If the analysis is positioned as an abstract report rather than a solution to a pressing business challenge, it will likely be ignored. However, if stakeholders are involved in shaping the questions, if the data is designed for cognitive ease (to be discussed in an upcoming post), and if the results are embedded directly into decision workflows, action becomes the natural outcome.

What is Choice Architecture and Why Does It Matter?

The concept of choice architecture has been highly influential in my thinking about how to overcome the last mile in analytics.

It was first introduced to me in the Deloitte research report, The Last-Mile Problem: How Data Science and Behavioral Science Can Work Together, back in 2015. However, the concept comes from behavioural economics and describes how the way choices are presented influences decision-making.

Richard Thaler, a Nobel Prize-winning economist, and Cass Sunstein introduced this idea to explain how small changes in decision environments can significantly impact outcomes—without restricting freedom of choice.¹ In the context of decision science and analytics, choice architecture helps solve the last mile problem by structuring the decision environment to increase the likelihood that stakeholders will act on insights.

Decision Scientists must go beyond delivering insights—they must act as Choice Architects. As Thaler and Sunstein explain, “A choice architect has the responsibility for organizing the context in which people make decisions.” By structuring the decision environment thoughtfully, Choice Architects increase the likelihood that decisions align with desired outcomes, addressing barriers such as indecision, bias, and information overload. In the context of decision science, this role is crucial for ensuring insights are acted upon and drive meaningful impact.

For example, instead of assuming people will engage with data rationally, we design systems that “nudge” them toward action by:

• Pre-Framing the decision context to create an expectation that insights will lead to action (e.g., asking, “If the data shows X, what will we do?” instead of “Should we act?”).
• Orienting insights around potential losses (e.g., “Without this change, we risk losing 10% of customers” rather than “We could improve retention by 10%”).
• Setting action-oriented defaults (e.g., pre-selecting the best recommendation rather than presenting multiple options).

By applying choice architecture in the pre-analysis phase, you ensure that data-driven decision-making is not just an option, but the easiest path forward.

The last mile isn’t a hurdle to overcome at the end—it’s a problem you solve at the very beginning.

Integrating Predictive and Prescriptive Insights to Drive Impactful Decisions

The Decision Science Pre-Analysis Framework is also heavily inspired by the objective of applying causal inference (as much as possible) and predictive and optimization methodologies (more broadly) to analytics.

This aligns with the principles outlined in From Meaningful Data Science to Impactful Decisions: The Importance of Being Causally Prescriptive by Victor S. Y. Lo and Dessislava A. Pachamanova. While achieving true empirical causality in every business scenario can be challenging, we can still design our analyses to come as close as possible by carefully structuring the process to account for potential cause-and-effect relationships.

As I have posted about previously, this approach ensures that analytics is not only descriptive but also predictive and prescriptive—empowering stakeholders with actionable, decision-ready insights.

As the authors explain:

“Successfully integrating all three levels of analytics in the context of business process improvement requires combining data science with decision science. Data-driven strategies that are not decision-driven strategies can be seriously flawed. From a methodological point of view, transitioning from data science to decision science requires (1) a holistic view of the process—from data to descriptive, predictive, and prescriptive analytics—and (2) generating the correct inputs through predictive analytics to inform prescriptive analytics.”²

The Decision Science Pre-Analysis Framework incorporates this principle by structuring analytics to flow seamlessly from descriptive (what happened) to predictive (what is likely to happen) to prescriptive (what should we do).

By prioritizing thoughtful scoping and analysis design upfront, the framework lays the groundwork to approximate causality in decision-making—i.e. in order to achieve a goal, we make a decision (or take an action) to cause a desirable outcome to happen—enabling teams to understand why outcomes occur and how best to intervene. This methodology strengthens the link between data and decisions, helping organizations drive impactful, data-informed actions.

Preview: The Decision Science Pre-Analysis Framework

Below is a preview of the six main pillars of the Decision Science Pre-Analysis Framework, designed to help teams solve the last-mile problem before the analysis ever begins.

While some of the pillars (3-6) involve communication and reinforcement during and after the analysis, it is critical that the process starts with the first three pillars. If these foundational steps—aligning with business priorities, pre-framing the decision context, and designing stakeholder buy-in—are not done correctly prior to analysis, it is highly likely that the analysis will miss its mark in terms of driving decisive action.

1. Aligning with Business Priorities

• Do stakeholders believe this analysis solves an urgent problem?
• Have we defined what success looks like?

2. Pre-Framing the Decision Context

• Do stakeholders expect to make a decision based on the insights?
• Are specific decisions clearly linked to measurable outcomes and the broader goals those outcomes are intended to achieve?

3. Designing Stakeholder Buy-In & Ownership

• Have we involved stakeholders in defining methodology & success criteria?
• Have we assigned a decision owner for this insight?

4. Structuring Data For Cognitive Ease

• Is the data structured for cognitive ease (clear, minimal, prescriptive, intuitive)?
• Have we tested the best visualization/storytelling method for the audience?

5. Embedding Insights into Decision Workflows

• Will this insight be embedded where stakeholders already make decisions?
• Have we set up automation, or alerts to reinforce action?

6. Setting-up Accountability & Reinforcement Mechanisms

• Have we tied this insight to KPIs?
• Is there a follow-up meeting scheduled to review implementation?

How This Framework Applies Choice Architecture

The pre-analysis framework above is designed to “nudge” stakeholders toward action by shaping their decision environment in ways that:

• Reduce friction in decision-making
• Mitigate cognitive biases that lead to inaction
• Make data-driven decisions the default choice

More on these topics in the weeks ahead, too!

Final Thought: Make Action the Default

The best analysis in the world is useless if it isn’t acted upon. By designing the pre-analysis phase with choice architecture in mind, you shift decision-making from passive data consumption to active implementation.

Over the next six weeks, I’ll be diving into each of these six pillars in more detail, exploring how to optimize each phase to ensure your insights don’t just inform—but drive real business impact.

Stay tuned.

Preview: if you would like to see a slightly more detailed draft outline of the Decision Sciences Pre-Analysis Framework (WIP), click through to the next page.

Pages: 1 2