Those who work with vast quantities of data know by now: It’s fast becoming the world’s most valuable resource. This fact has only accelerated in 2020’s remote reality, where even brick-and-mortar businesses that don’t typically focus on their digital footprint are scurrying to make their products available to buyers online. Krystal Lau, AB InBev’s Global Director of Product Data, reveals why it’s essential that each team understands how customer data is collected—and how it impacts the bottom line.
In this digital present, unlimited datasets are available. The data that powers the retail revolution and a working knowledge of how to leverage it properly becomes non-negotiable for all companies. As the Global Director of Product Data and Insights at Anheuser-Busch InBev, Krystal Lau is navigating this new world in real time as she and her team work to make it easier for ABI’s B2B clients to seamlessly order and receive product. Because personalized information is the fuel that powers the digital marketplace, it’s essential to establish a source of truth that helps every member of the team grasp where the data comes from, what triggers its collection, and how to analyze and apply those insights for the highest possible ROI. In this episode, she shares how they pull off such a tall order.
This is Segment’s Data Council series, where members share stories about using Segment to work with customer data within Enterprise companies. To make sure you don’t miss an episode, subscribe on iTunes, Spotify, or your favorite podcast player. You can also read a lightly edited transcript of the conversation below.
Takeaways:
Audio highlights:
Links:
Madelyn Mullen: I'm Madelyn Mullen, part of the Segment Product Marketing team. For this episode, I'm joined by Krystal Lau, Global Director of Product Data and Insights at Anheuser-Busch InBev. Thank you for joining us, Krystal.
Krystal Lau: Thanks for having me.
Madelyn: Krystal, what have you been focusing on at Anheuser-Busch InBev?
Krystal: The product that we're building is a B2B e-commerce application. If you think about ABI and our customers, they're primarily storefronts, bars, restaurants, or businesses that are consuming ABI products. Our platform is meant to make ordering much more seamless for those businesses.
Prior to this product existing, the way that they would order our products would be either making a call to a customer service center, or they would have a scheduled visit from a business development rep (BDR) who would visit their store for four minutes once a week. You'd have to be at your store during that visit to be sure that you could place your order. They would give you a delivery date, and you'd have to be at your store for that delivery date or else risk not receiving that order and not being able to pay for that inventory.
We wanted to build a product that could make that store owner experience better so that they could go browse products from our portfolio. They could place these orders whenever they had free time. They could schedule a delivery date. They could pay in the product itself. They would get messaging on when to expect their delivery, so they didn't have to just stay at their store for two days straight because the delivery truck was delayed.
It also allowed the business development reps — who previously were just running from store to store, pushing products, and making sure that orders were delivered — to serve as more strategic account reps by taking a look at what orders folks were making on the platform.
That’s the product that we're building. We started in the Dominican Republic, and we're kicking off in other markets that we work with in the world. My role here at ABI is related to everything being tracked on that product. [That includes] everything related to the data architecture, what events we're capturing, how those events tie into each other, and then, ultimately, how it's leveraging that data. It's making sure the data that we're capturing has use cases in the business, that folks know what the data means, that they know how to use it, and that they understand what it's saying. Also, we have key metrics for success that are defined and consistent from market to market. When they have questions around different areas of the product, or when they have questions around how the product is influencing their own offline strategies, they have a source of truth and someone who they can work with to make sure they're understanding the data the way that it's meant to be understood.
Madelyn: Krystal, when working with customer data, what are some of the relationships and roles involved?
Krystal: Customer data at Anheuser-Busch InBev means a lot of different things in the traditional company. Folks at ABI have looked at customer data in relation to customers calling our customer service centers or customers talking to our business development reps and asking for orders to be placed on their behalf.
But in my world, everything is related to digital customer data. On our product, every interaction that a customer can have with our platform — every click that they do, every product they add to the cart — is something that I am looking at day in and day out. It's become such a large part of the conversation, because for so long, everyone has been looking at invoice data or very analog offline data. To have the ability to see user actions in real-time has been game changing.
The folks who use our customer data range from team to team. We have product and engineering folks who care about, "How are users flowing through the product funnels?" and, "How are they using the different features that we built?" and, “How does that compare to other features that we have available?"
Versus, someone who's in the revenue management team and cares about, "What are users adding to the cart, but then not buying?" or, "What are users buying together?" Because that influences when you bundle products together in some form or fashion.
If I'm in the logistics team, and traditionally all I've had was insight into invoices placed and then the ultimate deliveries being made, I might want to understand the customer side of things, which is: How many of those orders and invoices were canceled from a customer's perspective versus how many did we cancel because of lack of inventory?
There are so many different applications for how we use the customer data, and every team at ABI is a customer of that customer data.
Madelyn: Krystal, if we were catching up over coffee, would you tell me a story about a time that you were using that data, and it wasn't quite making sense? It wasn't quite what you were expecting?
Krystal: At ABI, I started in October 2019. I think me joining was a signal that they wanted to understand more about that customer data. We were capturing a lot of data, but no one was deciphering what any of it meant. Every week when I first started, it felt like it was a new learning for myself and for everyone else as well.
Every week it was a new question! These were just a few:
Some of that data made sense, and some of that data did not make sense. In one example, we saw a ton of orders grow from the last eight months or so. From August to April, orders have only continued to grow every single month thereafter. We got the most delivery ratings submitted. After customers place an order, the next time they log into the application, they get one of those surveys where they have to rate that order in one to five stars.
Right when we first launched that functionality, we saw ratings come in left, right, and center. Everybody was giving us a ratings, so we had a huge sample to go off of. It turns out that as our orders continue to go up, the amount of delivery ratings we were receiving were going down, and we weren't entirely sure why that was.
There were a couple of theories: Maybe folks are ordering in an automatic fashion almost every single week, and they have a very similar experience every time they're doing it, so they can just opt-out of those delivery ratings if they needed to. There were other theories: Maybe they have some messaging fatigue in terms of getting all of these different in-app surveys as well as messages sent directly to their phone. Also, they were getting messages sent directly to themselves as they're browsing through products. There were also theories that, because we were oversaturating how many different ways we were communicating with them, they were just shutting us out and not giving us any feedback whatsoever.
Those are things that are still open, but it's always interesting to see what data doesn't make sense to understand a little bit more about how our users are responding to this very new product that we've launched in their market.
Check out the Segment Analytics Academy lesson ‘Take your product usability to the next level.’
Madelyn: Tell me a bit more about the "we." Who's been involved when you're looking at that type of rating discovery as well as other data questions?
Krystal: It's the product data team, so myself and a couple of other folks that report to me. It doesn't make sense for us to be looking at data in a silo. We're often partnering with logistics teams, folks that work at ABI, and the teams in the different markets. In the Dominican Republic, for example, teams are looking at your sales and inventory data that are talking to the different sales reps, customer service center folks, and the local team on the ground knows the ins and outs of the operations in those markets. What we do often is work closely with teams that have more of that context. The insights they have — since they're not in a silo — can inform something new rather than restate what people know anecdotally.
Madelyn: Since this is a new app, have you started to see some of these business relationships or logistics relationships change?
Krystal: Yeah, they've grown stronger since I've been here. I've only been at ABI for six months, but I already feel more connected to folks as time goes on.
There was a lot of appetite for data when I first joined. Everybody loved data, and everybody wanted to learn more about what users were doing on our application. That being said, there's one thing to get that insight and think of it as a cool finding. There's another thing to understand the value it provides to them and their day-to-day jobs.
In the example of logistics, everybody was very positive about the data that we were collecting and how it could be very complementary to the very analog data they were used to looking at on their end. It's been the best, happy marriage where we will get new data being tracked and new tracking implemented every sprint. Then with the release of that tracking, we can append new things to the reporting they have. Overall, we can inform or answer questions they previously had but could never get answers to. And we could even provide something nobody had an answer to but would incite more questions. With that, we could drive the strategy in a different way.
Logistics is one example, but there are so many teams where their understanding of our data stack has helped bridge a lot of the gap between them and myself, for sure. But it also is a bridge into if they can understand the data and how it's being collected, they can also understand the product, how a product is developed, and how the features that are digital in nature reflect on their offline strategies.
Madelyn: Are there some specifics you could share for the logistics team or any of the local teams?
Krystal: We recently started launching a couple of new products into the Dominican Republic. Prior to this, we had our portfolio of products that we had available to the Dominican Republic. We've started experimenting with launching a couple more.
And one question we got for day one after launch was their appetite for these new products. Very quickly — because of the real-time nature of our data and how we were collecting it —within an hour of launch, we could categorically say that "Yes, of course there is appetite."
Then the next question was, "How many people placing those orders are repeat-purchasing those products?" Then you go another week after folks are going through their standard buying cycle. And of course, you can see that people are religiously placing orders on these new products week after week. Then we can start to think about, "If people are enjoying this new SKU of products we've launched, does that tie into any of the older SKUs that we have?" And, “Are people ordering them in the same order or in very quick orders in succession?” The reason that's important is that if they're placing orders together in one order, then that can tie into how those placements are in the app itself.
It's very tactical, very product-oriented. But if users are placing orders right after each other and literally going through the app and placing one order per day — similar to how, if you don't have to worry about delivery costs or that you only get one scheduled visit from a BDR, you're more likely to place more orders in a short period of time, even though you'd only get invoiced once.
Basically, you're changing the behavior of how you're ordering. That can tell us — if folks are placing lots of successive orders, and there's a similarity in the products that they place in those orders within that week — there's something that can be done related to bundling products together in terms of making sure that inventory is available of these products together in combination, etc.
For the logistics team, it's more important for them to know that they can fulfill all of those orders, that they can get those delivered, and that there are no cancellations on those orders. For the revenue management team, since they're in charge or responsible for the pricing and discounting of those products and bundling those together, they care a lot about what that behavior is and how closely together folks are ordering stuff. They previously had data related to invoices, but it wasn’t so granular in terms of how quickly after placing one order did folks change their mind and want to add additional products. It's almost like a train of thought in how they're ordering, and a train of thought placing these orders with the expectation that they will all be delivered on the same day. We needed to bundle orders together accordingly and also make sure that we can fulfill those accordingly.
Madelyn: You've definitely helped that team look at data differently, as they're getting all that extra information. How have you looked at data differently from your onboarding in the past six months?
Krystal: Prior to being at ABI, I had been so used to looking at data for the media world. I came from Time Inc., and I was so used to looking at very large websites where it was mostly anonymous users. We would look at traffic trends on top content and use that to derive trends in the zeitgeist: what people are reading about, what people care about, what content they're looking at, and how that's driving cultural narratives.
I still refer to all of the things I learned back then in my day-to-day life, but coming into ABI, everything was very different. We had a smaller subset of users and global traffic. Every click is quantifiable in terms of revenue. For every eyeball that you have, you can literally see the path to completing an order. The applications of using this ABI data were totally different from me looking at the top content on magazine websites back in the Time Inc. world.
It changed the way [I would interact with] the standard toolkit I would use for analysis. Back in my Time Inc. days, we would look at top content trends and append them with perhaps some search console data, SEO data, or even Wikipedia data just to see how the trends were performing. What we are looking at in the ABI world is our own data: how it appended to other datasets that we have internally and how that can enrich the data that we're collecting through Segment and have one collective view of how the business is doing not just digitally, but holistically.
Madelyn: Bringing up Segment, what roles had you been working with Segment at Time Inc. and now at Anheuser-Busch InBev?
Krystal: At Time Inc., my primary role, on a very small data team, was being the ultimate arbiter of insights and reporting. Similar to the role that I have now, but it was me being that source of truth of, "What does the data mean?” and, “How is what we're collecting tying into the business questions that folks around the organization have?” At Time Inc., that meant if I'm an editor, “Why do I care what your Segment data stack is?” And the answer there is, “Well, we're collecting all of this traffic and the appropriate metadata related to your pages so that you can understand how the content that you are specifically writing performs, how it relates to other content on the site, how it relates to other content in the portfolio.” And then, “Is it moving the needle?”
Finding ways to relate the technical aspects of what we're doing with the business questions that people were asking — especially when a lot of those business folks were non-technical — was my primary role.
Coming into ABI, it's similar. I was tasked with being the arbiter of the insights and the data, but then also making sure that the analytics implementation of that tracking was executed in a way that would make that analysis simple, easy and scalable. Whereas at Time Inc., I was more of an insights person with visibility into the implementation, at ABI I became the implementation person but with all of the insights related questions helping feed that priority.
Source: Unsplash @ryoji__iwata
Madelyn: How did that feel going from the team providing the source of truth to the team now creating the source of truth from collecting all that information using Segment?
Krystal: It felt great. I'm not going to lie, it was very daunting. When you're the person responsible for the insights but not responsible for the implementation, there's a level of, "The data is what it is, and we can look at it this way because these are the limitations." When you become that person creating the data and the source of truth, then you have a lot more flexibility in making sure that all of the use cases are accounted for. You also have the responsibility to speak to why things cannot be tracked in a way that users want, and then finding alternatives that can still satisfy what their ultimate goals are.
Madelyn: How do you feel that data has helped you advance and influence your career?
Krystal: When I first started in media, I was not in a data role. I was leveraging another team's data to create a growth strategy for certain sites. It was nice to be able to look at data and come up with strategic decisions around it, but it's way more powerful to know exactly where your data is coming from and to know what triggers what, what means what. Because if you know all of that, then you broaden your toolkit into what you can possibly do with that data — and not just in terms of analyses but in terms of the applications of those analyses or the applications of that data itself.
I enjoyed moving into a more strategic data role where it's a different kind of strategy, but it unlocks a lot of capabilities that teams didn't even know were possible. It shed some light on tons of business questions where if I wasn't in the data world, I would have no idea about them. Being so central within the data world has allowed me to have the most souped-up onboarding experience I've ever had in terms of understanding business questions, how metrics can apply to different teams around the organization. As in the same data point could have many applications to different teams, and each team would leverage that data point in different ways. That's something that I wouldn't have visibility into if I wasn't in the data world.
Madelyn: How did you approach starting this new role in the data world where you're actually starting to implement Segment and some of the collecting of data?
Krystal: When I was interviewing at ABI, I met with a huge panel of people. I met maybe 10 or 15 people. Through all of those conversations, I got a pretty good sense of the current status of the data architecture, what tools were in place, what was the general data structure, what was the company structure, and how did the different groups within the company leverage that data.
Then the first month after me joining was very much about taking the time talking to people, overturning every stone, looking under the hood, trying to get a sense of how things were done, how data points were tracked, how different teams built their own self-service reporting and why — and if it made sense for the goals that they had.
I was pretty lucky in that I had a good couple of months after starting to dive so deeply into all of those questions. Because a couple of months into me joining, it was hitting the ground running in terms of now: How do we change? How do we evolve? How do we optimize the current processes? How do we make sure that implementing Segment is seamless? How do we make sure that we have a good QA process? How do we make sure that folks in the organization know what Segment is and how it relates to them? Do they need to know exactly what every component is? Probably not, but I found that it was helpful especially after me coming in and learning about all of them in their roles. Them ultimately learning about me and my role and how the data architecture would impact them has allowed for very strong relationship-building, and to be honest, just a ton of context on the business side that I don't think I would have had if I didn't start building those relationships.
Madelyn: When you were picking up on that context, were there any surprises you uncovered during your onboarding?
Krystal: At ABI, because it was such a new industry for me, everything was a learning — more on the context side of things and seeing the different applications of how our data was being used, which is always very interesting.
Some teams went for it in terms of leveraging our data for modeling and propensity scoring across our user base to provide recommendations. Those were very advanced applications. There are some teams that didn't know all of the different properties that we were tracking. Them using the data was also very interesting, because misinterpretations abound.
What was refreshing was after I met with folks across different teams, everybody was very open to optimizing their processes. They wanted help. They wanted someone who knew that data to help provide feedback on applications of that data. In terms of context and in terms of seeing the applications of our data, some of them were imperfect. But everybody was very open to changing the way they were doing it if they needed to. And some of them didn't. Some of them were great.
Madelyn: When you started using the Segment application, was it what you expected? Were there any surprises?
Krystal: When I was at Time Inc., our data team was so small — we were three people. Even though I was never the primary person going into the Segment UI and changing the configuration, I always knew how those configurations were set up, and I was able to look under the hood if I needed to. I think that helped me a lot just knowing what to expect coming into my new role at ABI, where I hadn't primarily had to deal with the Segment configuration before but now I was going to be doing that.
My comfort on that topic was eased because of that prior overview at Time Inc. Then in general, I like to read a good document over a glass of wine. Even though that might be nerdy.
Madelyn: Any documents you've read recently for Segment or other apps that you recommend?
Krystal: I think I've memorized the Segment e-commerce spec, frontwards and backwards. The Personas documentation is top of mind right now. It's something that we've started looking into over here.
Madelyn: What are some of the projects or use cases you might want to use Personas (our user profile and audience tool) for?
Krystal: There is a lot of testing that we want to do on our platform. There's a lot of targeting in terms of messaging that we want to do in our product. We leverage messaging tools to be able to reach our customers. It all works as expected. We are able to reach all of the users on our platform, and it's great.
What we've found is eight months into our product being live, we're messaging folks too much. We're reaching them too much to the point where they are not converting at as high of a rate as they previously did. What we're starting to look into now — and what Personas can be very helpful for — is starting to segment those users in terms of SKUs that they've previously purchased, whether they're repeat buyers or if they are buying a new SKU for the first time.
[There are all kinds of] different actions that they could be taking if they clicked on our product to add to the cart, but then they ultimately abandoned their cart. There are things that we could use Segment Personas to help make those audiences smaller so that we're not reaching the same people many times a day or a week or a month, and we’re using our resources more efficiently, wisely, and effectively.
Madelyn: Krystal to wrap up this discussion, what are three takeaways on Segment’s first or second impressions you'd like to leave with the listeners?
Krystal: Teams are not always going to speak the language that you speak. They aren't always going to understand what Segment is, what tracking is, and what our tracking schemas are. And that's fine. But it's always important to understand what is important to them and how data can be used in their roles. Even if they may or may not be using some data, it's important to know their ultimate goals.
That helps as you're developing your data stack and your data architecture. It helps you make sure you're not just tracking things for the sake of tracking them, but you're always tying it back to a use case and defining an application for what you're building. Therefore, it just makes it much more powerful, much more scalable, and it helps build the relationships and the buy-in from the business once you are able to translate the technical [aspect] of what you're doing to how it applies to their lives.
The second takeaway is just understanding generally that the same data point can have multiple use cases. Similar to the first point: doing your due diligence, talking to different people in the organization, understanding what their different goals are.
What you also will find is that different teams will leverage the same data point but just in different ways. With your data architecture, it allows you to collect fewer data points but more robust data points. It makes every piece of work that you're doing — if you can bring in all of these differing use cases and opinions and have them all spoken for in a very simple and elegant data point — easier to analyze around how different areas of the business impact each other or the whole altogether. It makes it easier to help translate from one team to the other in some instances.
The third is don't underestimate what impact a five-minute overview of your data stack can have when building a relationship with different teams in the business. Once folks understand the data infrastructure and how it ties into not just their team but other teams in the organization, nine times out of 10, they want to get on board. They want to ask questions. They want to see if the data that you have can solve their questions they've never been able to answer. And if you can start chipping away at some of those very difficult questions, then you've got someone beating the drum in a different area of the business. It makes it easier to get more folks on board and more folks asking those questions and providing that context and helping you understand how your data can be enriched further.
Madelyn: Krystal, thank you for sharing with us the power of data.
Krystal: Thank you for having me. This was fun.