Stakeholder Buy-In for Data Science: Home Depot, Mobilewalla, MLB
At COLLIDE 2023 in Atlanta, a panel of data consumers from Home Depot, Mobilewalla and Major League Baseball explained what makes data science useful to them. They described aligning on business goals, translating metrics into plain language, and building relationships between analysts and decision-makers.
By The AI Enterprise Magazine Team4 min read
Key takeaways
- Align on the problem first. Home Depot's search team sits with data scientists to agree on the business goal, which makes buy-in more natural.
- Collect only actionable data. Mobilewalla's Laurie Hood focuses on data marketers can use, such as share gained or lost.
- Build an exchange rate for metrics. Abdo mapped a technical ranking metric to conversion and order value so leaders could follow.
- Do not assume stakeholders are non-technical. Some want to explore the data themselves.
- Relationships carry the culture. In baseball, analysts in the clubhouse build trust even when players do not follow every number.
At COLLIDE 2023 in Atlanta, a panel of data consumers, not data producers, explained what makes data science work for them. "The Voice of the People: Stakeholder Perspectives on Data Science" brought together Kevin Abdo, Senior Manager of Product Management for search at homedepot.com; Laurie Hood, Chief Marketing Officer at Mobilewalla; and Matt Wieters, a retired 12-year Major League Baseball player and Georgia Tech assistant baseball coach. Anqi Zou, SVP and Innovation Leader at Truist, moderated.
Their shared message was that alignment, a common language and relationships matter as much as the models. Panelists said their views were their own and not their employers'.
Why does stakeholder buy-in matter for data science?
Abdo said buy-in at Home Depot hinges on two things: data scientists understanding the underlying business problem, and the team agreeing on the business goals. His product team sits with the data science team to discuss what the customer is saying, both quantitatively and qualitatively. With that alignment, he said, buy-in "happens more organically and naturally."
Hood said that at any company size the approach has to be collaborative. She advised thinking broadly about stakeholders, since a project may need finance for funding or legal for review, and explaining it in terms everyone can follow.
Wieters described the same challenge in baseball. Analytics first changed how front offices evaluate players, then how coaches strategize, and now, he said, the task is getting players to buy into training around the data.
How do stakeholders interpret data outputs?
Abdo said his team starts with strong hypotheses and usually A/B tests them. He also tries to influence a model's inputs by sharing customer insights with data scientists, then combines the quantitative results with that context into a story non-technical leaders can follow.
Hood warned against collecting data just because it is possible. Her marketing customers, she said, want to know where they are gaining or losing share, and from whom. "I don't need to collect what I can't use," she said of her own data.
Wieters said baseball is reactionary, with less than a second to decide, so he favors simulations that give players more reps.
What challenges slow adoption of analytics?
Abdo cited basic inconsistency, such as team members defining conversion or click-through rate differently, and the risk of analysis paralysis when a model works for one customer segment and not another.
Hood said the hardest moment is when data tells a senior leader something they do not want to hear. She agreed that at some point a team has to decide with the information at hand, while continuing to test.
Wieters noted that much of baseball is personal and relational, and a player may trust what he has seen over a number. Zou added that some organizations have moved from the term data science toward decision science, because the goal is making decisions.
How can data teams communicate better?
Abdo gave a concrete example. Data scientists told him about a jump in normalized discounted cumulative gain, a search ranking metric, and he had no idea what it meant. His team looked back at offline tests and built what he called an "exchange rate" between that metric and measures such as conversion, revenue per visitor and average order value. Now he can ballpark the likely business impact of an offline result, though not precisely.
Hood said larger organizations often benefit from a strategy or analytics group in the middle that translates between marketers and data scientists.
Zou cautioned against assuming stakeholders are non-technical. Some want to explore the data themselves, she said, and sharing it can lead to a healthier discussion than showing only polished dashboards.
How does data science change company culture?
Abdo said his team worked to dismantle silos between data science and product, and ran a "road show" to demystify the search engine, once a black box. Merchandising and other partners now seek the team out, he said.
Hood said Mobilewalla's CEO holds a PhD in data science, so little cultural change was needed. Wieters said analysts are increasingly present in clubhouses, and that even where players and analysts do not fully understand each other, willingness to talk is the shift that counts.
Closing advice for data teams
Asked for one piece of advice, Abdo said to understand the business problem and align on goals. Hood chose communication and empathy. Wieters chose relationships. Zou summed up the session in three words: relationship, communication and alignment.
Where to continue the conversation
Sessions like this continue at COLLIDE 2027, the largest AI conference in the Southern U.S. with more than 1,000 enterprise data and AI leaders, on Wednesday, October 13, 2027, at the Sandy Springs Performing Arts Center in Atlanta. See /events/collide/. Registration is coming soon. Free live sessions are also on the 2027 webinar schedule.
Frequently asked questions
How do you get stakeholder buy-in for a data science project?
Panelists said to align on the business problem and goals up front, involve a broad set of stakeholders and explain the work in plain terms.
How can data scientists explain technical metrics to business leaders?
Kevin Abdo described mapping an offline search metric to familiar measures like conversion and average order value, creating an approximate exchange rate.
What is the biggest challenge in adopting data insights?
Panelists cited inconsistent metric definitions, analysis paralysis, and data that tells senior leaders something they do not want to hear.
How is analytics changing baseball?
Matt Wieters said analytics moved from front offices to coaches and now to player training, where the challenge is getting players to buy in.
This article is based on remarks by Kevin Abdo, Senior Manager, Product Management, homedepot.com; Laurie Hood, Chief Marketing Officer, Mobilewalla; Matt Wieters, retired MLB All-Star and Georgia Tech assistant baseball coach; moderator Anqi Zou, SVP and Innovation Leader, Truist at COLLIDE Data + AI Conference 2023 on October 3, 2023. More coverage: COLLIDE AI Conference.
Filed under: People, Culture & Leadership, AI Strategy & ROI
