“Don’t Wait for Perfect Data, Start, Then Improve,” says Sam Wong, Senior Director of Data and AI at Mark Anthony Group

December 15, 2025

Sam Wong, Senior Director of Data and AI at the Mark Anthony Group, explains how a business-first AI incubation program can turn experiments into production value. He outlines a five-part framework, the importance of executive and business sponsorship, partnering with a vendor ecosystem, and prioritizing use cases by value versus effort. Wong discusses why you shouldn’t wait for perfect data, how AI projects catalyze data governance and quality, and how a blended federated/centralized operating model scales delivery in a mid‑market company. He also tackles change management and job-loss fears, positioning GenAI as augmented intelligence. 

• What does it take to build an incubator that learns fast and delivers real value?  
• When is “good enough” data enough—and how can AI expose and improve the gaps?  
• Which operating model and governance practices drive adoption without slowing execution?  

Watch the interview 

Read the podcast as an interview  

(The interview was shortened and edited using ChatGPT

Mark Kohout: Hello and welcome to the Adastra podcast. I’m Mark Kohout, and I lead the governance practice at Adastra North America, a data and AI consultancy powered by the cloud. I’m joined in our Toronto studio by Sam Wong, Senior Director of Data and AI at the Mark Anthony Group (MAG). 

If you don’t know MAG, it’s an entrepreneurial producer of beverages that pioneered new categories with iconic brands like White Claw Hard Seltzer, Mike’s Hard Lemonade, and most recently, Más+ by Messi, a next‑generation hydration drink in partnership with football legend Lionel Messi. MAG grew from a one‑person wine import business into an international drinks company through a spirit of relentless innovation. 

We’ve asked Sam to talk about MAG’s AI incubation, what you were looking to achieve and how you approached it. 

Sam Wong: Thanks, Mark. When we talk about data and AI, we don’t just focus on the technology stack, tools, and platforms. It’s about what we’re trying to solve, what business outcomes we’re driving, and what value we deliver for stakeholders. Innovation isn’t tech for tech’s sake; it’s how tech enables us to improve what we have at Mark Anthony. 

Mark Kohout: Where did the incubator idea come from, and how did leadership support it? 

Sam Wong: It started with discussions with our CEO and CIO about the explosion of AI. I proposed an incubator, a safe space to explore the technology and how it enables the business, without forcing timelines or deadlines. That let us experiment while targeting strategic benefits like revenue growth, productivity, and improved customer experience, moving promising ideas into test and pilot. 

Mark Kohout: You say “learn fast,” not “fail fast.” What did it take to mobilize the program? 

Sam Wong: We developed five components: 

1. Establish the program and mission. 
2. Intake and curation. 
3. Build and execute. 
4. Identify lessons and improve. 
5. Demonstrate and market. 

Showcasing what you’ve done isn’t about ego; it inspires what’s possible and triggers more ideation. 

Mark Kohout: What makes an AI incubator successful, and what do leaders often overlook? 

Sam Wong: Have a mission statement: clarity on purpose and goals. Get executive sponsorship and a strong business sponsor who ensures the work leads to value and helps make it real. The most successful initiatives had strong collaboration and room to pivot based on experimentation, iteration, and feedback. We adjusted when blockers appeared and let solutions evolve into something ready for implementation. 

Mark Kohout: Did vendors or partners play a role? 

Sam Wong: Yes. Mid‑market organizations need to be pragmatic. We’re fortunate to have a strong vendor ecosystem, and our brand helped us gain sponsorship, access to resources, and sometimes early tech. Treat vendors as partners; you get more intellectual resources and horsepower, learn from those ahead of you, and share back with those behind you. 

Mark Kohout: How did you prioritize your AI backlog? 

Sam Wong: Use a matrix of business value versus effort. Aim for high impact aligned to strategy with low effort. Then balance value, time to market, and the bandwidth of your business sponsor. Flexibility matters; without forced deadlines, you avoid cutting corners just to get a product out. Let things grow organically. 

Mark Kohout: Did you have to look closely at data management and readiness? 

Sam Wong: Absolutely. “Garbage in, garbage out” is true, but don’t wait for perfection. There’s still judgment and curation. AI projects can kick‑start your data management journey (data quality, governance, master data) and trigger buy‑in to take ownership and improve data. Treat AI and data as a journey and a program, not a project. Allow growth, iteration, improvement, and treat it like a product as use cases evolve. 

Mark Kohout: Moving from experimentation to production, how did your approach perform, and what pushed pilots over the line? 

Sam Wong: Our batting average was high because of collaboration and the willingness to pivot. Work feeds into other work; one idea can lead to another that goes to production. We were conscious about privacy, enterprise architecture, and infrastructure, keeping those groups involved and guardrails in place. Pilots were key: see it in action with a smaller cohort, get feedback, build excitement, and incorporate ideas. Drop the ego, ideas can come from everywhere, and adoption goes up. 

Mark Kohout: It sounds like governance was built in with a multidisciplinary team. 

Sam Wong: I had oversight over multiple data pillars: AI, data engineering, and data management, so we avoided siloed builds or tech that wouldn’t meet governance standards. Seeing how everything connects lets you fix one area to make another easier. It’s a bit of chess; think beyond the immediate incubator. 

Mark Kohout: How did you staff and fund the work without a full‑time R&D bench? 

Sam Wong: We blend a federated model with a centralized model. Customer‑facing leaders sit with stakeholders, understand the business and data. Centralized “factory” teams (AI/ML engineering, data engineering, analytics engineering, data management) scale up or down as demand changes. Think homebuilding: you coordinate specialists when needed while keeping core leadership and architecture in place. 

Mark Kohout: One piece of advice for leaders getting started with GenAI? 

Sam Wong: Address job‑loss fears upfront. The intent is to supercharge teams: more powers and capabilities. I call it augmented intelligence: AI and people working together. Start change management early, position AI as an enabler, and you’ll get more engagement and better outcomes. 

Mark Kohout: And how do you bring skeptics on board? 

Sam Wong: Spend time with them. Workshop pain points and opportunities; don’t start with tech or data. Deconstruct the problem, show how data practices and AI capabilities meet it, and outline the ROI. Once you agree on what’s important to solve, the solution part is easier. 

Mark Kohout: Sam, this has been a fascinating conversation. Thanks for joining us, and thanks to our audience for listening. Stay tuned for more on data and analytics, and please like and subscribe. 

Listen to the interview