The en masse shift to online shopping that transpired during and post-COVID appears as if it’s destined to last. According to the Brookings Institute, one third of US adults have used delivery apps to order from a restaurant or store in the last year.
In light of this growth, AI-enabled product recommendation engines naturally appeal to online/hybrid retailers. Product recommendation engines can increase cart value, enhance customer engagement, and generate repeat sales.
As the largest retailer in the world by revenue, Walmart has a massive footprint in the online retail space and, by necessity, a significant interest in product recommendation systems.
Emerj was fortunate to speak with someone who has worked with Walmart on a large-scale, related machine learning project. Dr. Charles Martin is a preeminent figure in the world of AI and machine learning. Charles is the founder of Calculation Consulting, an AI consultancy and software development firm based in Silicon Valley.
Charles gave an insider’s perspective to Emerj CEO Daniel Faggella on the AI in Business podcast, focusing on what working on a complex technical project within such a large company takes to find success in two key areas:
- What to look for in a project champion: Finding a knowledgeable, experienced domain expert that possesses the intangibles of a superior work ethic, trustworthiness, and reliability can be the key success factor in determining project success.
- How to put the right people around you: Applying selection criteria similar to that of the project champion in picking project workers.
Listen the full episode below:
Guest: Dr. Charles Martin, Founder, Calculation Consulting
Expertise: Data science, machine learning, deep learning, natural language processing, consulting.
Brief Recognition: Dr. Martin is an early pioneer in machine learning, an AI specialist, and an engineer. He has taken lead roles in AI design and implementation for marquee name clients, including GoDaddy, Walmart, Google-acquired Aardvark, and eBay. Dr. Martin holds a Ph.D. in Theoretical Chemical Physics from the University of Chicago. He did his postdoctoral work in neural networks at the University of Illinois Urbana-Champaign.
Project Champion Criteria
Charles is no stranger to complex implementations with global companies, having worked on critical enterprise deployments with names like GoDaddy, eBay, Google-acquired Aardvark, and in his current role at Walmart.
In this episode, Dr. Martin hones in on a single use case at Walmart that involves improving the retailer’s product search and recommendations engine in different languages and geo-locations. Charles also discusses some of the challenges and strategies for overcoming these challenges.
Walmart’s international online retail platform was experiencing a massive surge in orders during and post-COVID. The retailer sought to employ AI and machine learning to assist the company in expanding its international market presence by improving the online shopping experience via an enhanced product recommendations engine.
Dr. Martin credits a “champion” within the organization as the core factor in achieving project success. In the case he cites, it was an employee within Walmart with several years of experience in search. The individual understood search to be a complicated problem requiring an innovative solution. Critically, this champion had a depth of knowledge on technical implementations within the search domain.
The champion must also be a fighter in that they must be willing to go to bat, if necessary, to defend the project’s rationale. The sponsor must also be ready to combat the litany of alternative solutions others within and outside the organization offers.
“Whenever you’re doing a project in a complex field, you have to have some experience. You have to know the industry. The other thing is that in a big organization, there are a lot of people with a lot of different ideas, and you have to pick your battles. And there’s a point where you get to say, ‘Look, this isn’t going to happen,” and we have to pick the battle. And you have to fight it.”
– Market People Operations Lead at Walmart, Dr. Charles Martin
The primary–though certainly not the sole–challenge of implementing a technical solution for an international market was not the language barrier, as one would assume, but market complexity.
Different markets behave differently, and it is necessary to examine each market and decipher and account for their dissimilarities. Accomplishing this requires intelligent use of the tools and knowledge available at one’s disposal:
“You have to understand what the components and processes are in an AI system because you have data collection. You have to go through the data in something like this. You have to get it running in production. You have to have a semi-automated system. You have to be able to run A/B tests. You have to understand statistical analysis. A number of components go in, and you have to understand there are different sub-components which interact with each other.”
– Market People Operations Lead at Walmart, Dr. Charles Martin
Project Worker Criteria
When implementing an AI initiative, it is imperative to bring in the right people; this involves reaching out to individuals with the requisite AI-related knowledge and experience.
Charles’ company was able to build and implement an end-to-end, proven machine-learning solution within a brief timeframe that met and likely exceeded client requirements.
The people you bring on must also embody the intangibles of a solid work ethic and a willingness to tough out the numerous challenges that will arise.
Regarding other success factors, Charles articulates the importance of experience in solving complicated problems. As a solutions provider, one can not overvalue this asset. Dr. Martin elaborates on what his extensive experience in the AI consulting space has afforded him in terms of surveying the project landscape:
“For me, what I bring to the table is that I’ve just worked in so many enterprises, and I know what to expect and what the barriers are going to be,” Dr. Martin tells Emerj. “So I know what the asks are early on: What are the resources we need? What don’t we need? What do you not want to do?”
Understanding specific client requirements and building a usable product as quickly as possible is highly advantageous. Many companies either expect or strongly want early wins– initial measurable impacts, financial and otherwise, from the solution being applied.
Demonstrating early value to stakeholders is a critical element to project success. In a recent podcast episode, Gero Gunkel of Zurich Insurance shared his take:
“It’s really important to show initial benefit, initial value, to your stakeholders, and also to your users, so they’re willing to continue to give feedback, and they’re willing to go with you on this change journey … And once we got some initial success … we started to build on it with new use cases.”
– Market People Operations Lead at Walmart, Dr. Charles Martin
You need to pick your customers wisely; more specifically, try to select those who grant a certain level of autonomy.
The same concept of autonomy applies to employees within large organizations. Problem solvers must be given the freedom to play around, investigate, and try different things.
The challenge is determining the degree of autonomy to grant, which may explain some organizations’ refusal to bestow the level of independence needed to see a successful project through.
“Sometimes organizations are the opposite. There’s almost no autonomy and no trust. But you have to, if you’re going to work in this kind of environment, be willing to take on that responsibility and say, ‘I’m gonna try to do what’s not just what’s best for me, as an employee, but what’s best for the project.”
– Market People Operations Lead at Walmart, Dr. Charles Martin
A champion will no doubt have a high level of autonomy. Here, Charles provides a list of related intangibles:
- Reliability
- Competency
- Trustworthiness
- Selflessness
For some business leaders, screening a potential champion using pre-defined criteria may be in order. For others, finding such an individual may be a matter of “feel”; looking for the one person who demonstrates a strong belief in the project, can intuit the people needed to make it successful, and displays the AI fluency and domain knowledge necessary.
“That’s very important in any organization I work with: you have to want to be an AI company, and you have to want to do it,” Dr. Martin tells Emerj. “If you don’t want to be an AI company, you’re not going to be. And if you want to be one, nothing’s going to stop you.”
