CMO Confidential Podcast - Detailed Summary
Episode: Dr. Joel Shapiro | Northwestern | The Grocery Prediction Case - It's Not Just About the Data
Release Date: May 20, 2025
Host: Mike Linton
Guest: Dr. Joel Shapiro, Professor at the Kellogg School of Business, Northwestern University
Introduction
In this episode of CMO Confidential, host Mike Linton engages in an insightful conversation with Dr. Joel Shapiro, a seasoned expert in decision science and data analytics. Dr. Shapiro, with a robust background encompassing public policy, business analytics, and law, delves into the intricacies of leveraging data to solve complex business problems. The focal point of their discussion is the Euro Grocer Case, a compelling study that underscores the challenges and nuances of data-driven decision-making in the retail sector.
Understanding Data and Decision Sciences
Grounding the Conversation
Dr. Shapiro begins by elucidating his journey from public policy to business, emphasizing the dynamic nature of business environments where data-driven decisions can rapidly impact outcomes. He articulates the distinction between large-scale "home run" ideas and the incremental, data-informed decisions that aggregate to significant improvements over time.
"I love the fact when data helps us make little decisions better on a daily, hourly, weekly basis because those things can aggregate pretty quickly and I like that pace."
[03:01] Dr. Joel Shapiro
Marketing vs. Other Business Functions
The discussion transitions to the comparative maturity of data utilization across business functions. Dr. Shapiro highlights marketing and operations as the most advanced in data-driven practices, contrasting them with sales, which he observes as less receptive to data interventions.
"Marketing really mature. Sales, not so much."
[05:08] Dr. Joel Shapiro
The Euro Grocer Case: A Deep Dive
Case Overview
Euro Grocer, a pseudonym for a major European grocery chain, faced an $8 billion annual sales landscape alongside a $250 million problem: inaccurate inventory ordering. This discrepancy led to both overstocking, resulting in spoilage, and understocking, causing lost sales.
Data Science Intervention
To address this, Euro Grocer enlisted a top-tier data science consultancy to predict category-wise demand across their stores. A pilot test revealed that the predictive model could potentially save $106 million annually by optimizing inventory levels.
"If I told you that you could get six and a half dollars back for every dollar that you spend, that's going to get your attention."
[11:42] Dr. Joel Shapiro
Board Resistance and Underlying Issues
Despite the promising pilot results, Euro Grocer's board declined the $16 million annual investment required to scale the solution. Dr. Shapiro identifies three critical factors contributing to this refusal:
-
Organizational Misalignment:
- The predictive model suggested precise inventory allocations, but Euro Grocer's existing supplier contracts and logistics systems lacked the flexibility to implement these recommendations effectively.
"So you can tell a grocery store that I can predict exactly how many avocados people are going to want to buy next week... you can't take full advantage of that, you know, predictive accuracy."
[13:14] Dr. Joel Shapiro -
Model Transparency:
- The consultancy employed a neural network model, prized for its accuracy but notorious for its lack of interpretability. Decision-makers struggled to understand the rationale behind specific inventory recommendations, leading to distrust.
"When somebody in a decision-making position says to you, explain to me why we should do this, when your only response is, because the model tells me to. Oof."
[15:57] Dr. Joel Shapiro -
Data Trust Issues:
- Errors in the initial data set, such as incorrect records of sales promotions, eroded trust among store managers. Even minimal data inaccuracies can significantly impact stakeholders' confidence in the data-driven solutions.
"Once people lose faith or trust in the data, it's really hard to get it back."
[16:26] Dr. Joel Shapiro
These factors collectively underscored the complex interplay between data accuracy, model transparency, and organizational readiness, ultimately preventing the adoption of a seemingly lucrative data solution.
Asymmetrical Risk in Data-Driven Decisions
Understanding Asymmetrical Risk
Dr. Shapiro expands the conversation to the concept of asymmetrical risk, using a case from the Illinois Department of Child and Family Services. A predictive model intended to identify at-risk children failed catastrophically by not flagging two children who tragically lost their lives, leading to the abandonment of the tool despite its broad effectiveness.
"There is a false positive and a false negative, but the consequences of one bring one terrible outcome to the kids."
[27:09] Dr. Joel Shapiro
Lessons on Managing Asymmetrical Risks:
-
Cost of Errors:
- Business leaders must thoroughly understand and weigh the consequences of false positives and false negatives to make informed decisions that align with their risk tolerance.
"Decision makers need to understand deeply the cost of being wrong, the benefits of being right."
[28:36] Dr. Joel Shapiro -
Balancing Model Precision and Organizational Capacities:
- Predictive accuracy must be matched with the organization's ability to act on insights, ensuring that data-driven recommendations are feasible and sustainable within existing operational frameworks.
Best Practices for Data-Driven Leadership
Building Trust Through Transparency and Communication
Dr. Shapiro emphasizes the pivotal role of trust in successful data initiatives. Establishing clear expectations and maintaining transparency about data limitations are crucial for fostering stakeholder confidence.
"If you're going to be successful with data, you got to set the right expectations and you got to be transparent with things when they are imperfect."
[22:46] Dr. Joel Shapiro
Effective Communication Strategies:
-
Know Your Audience:
- Tailor communications to address the specific concerns and interests of different stakeholders, ensuring that data insights are relevant and actionable.
"Know exactly what the purpose of your communication is... know what your audience cares about."
[22:51] Dr. Joel Shapiro -
Purpose-Driven Messaging:
- Clearly define the objectives of your data presentations, whether it's to persuade, inform, or garner support, to enhance the impact of your message.
Cultivating Data Leadership:
-
Dr. Shapiro advocates for the development of data leadership skills, urging data teams to collaborate closely with business units and focus on solving problems rather than dictating solutions.
"Practical advice is for the business teams maybe be a little bit more tolerant to the data teams... data teams to be really good at thinking more deliberately about how to influence business people."
[34:43] Dr. Joel Shapiro
The Role of AI and Future Implications
AI and Transparency Challenges
The conversation touches on the burgeoning role of AI in decision-making, highlighting the ongoing struggle between model complexity and the need for transparency. Dr. Shapiro expresses concern over the potential for over-reliance on large language models (LLMs) without adequate human oversight.
"AI has some adult supervision, you would feel comfortable."
[19:53] Dr. Joel Shapiro
Ensuring Responsible AI Use:
- Emphasizing the importance of human accountability and deliberate oversight to prevent misuse of AI-generated insights, particularly in high-stakes environments.
Final Takeaways
Key Insights for Business Leaders:
-
Data Doesn't Make Decisions:
- Ultimately, human decision-makers are responsible for interpreting and acting on data insights. Ensuring that data is meaningful and actionable is paramount.
"Data doesn't make decisions, People make decisions. And data needs to be helpful to them."
[31:57] Dr. Joel Shapiro -
Trust and Transparency:
- Building and maintaining trust through transparent communication and setting realistic expectations about data capabilities and limitations.
-
Aligning Data Models with Organizational Capabilities:
- Ensuring that predictive models are compatible with the organization's operational flexibility and strategic objectives.
Practical Advice:
- Foster a collaborative culture where data teams and business units work synergistically.
- Invest in data leadership training to bridge communication gaps and enhance the influence of data-driven insights.
- Always consider the human element in data interventions, recognizing that trust and understanding are as critical as technical accuracy.
Closing Remarks
Mike Linton concludes the episode by reiterating the essential lessons from the Euro Grocer Case and Dr. Shapiro's expertise. Listeners are encouraged to integrate these insights into their own data-driven strategies, ensuring that robust analytics are complemented by effective leadership and communication.
"Whatever else we discuss on other shows, like artificial intelligence... what's important is to remember the human element in data-driven decisions."
[35:10] Mike Linton
Stay Connected:
For more episodes and insights, subscribe to CMO Confidential on Spotify, Apple Podcasts, YouTube, and the I Hear Everything network.