AB InBev brews up greater insight into changing customer trends

Profile picture for user jtwentyman By Jessica Twentyman January 7, 2019
A new enterprise data hub  brings together a wealth of data on customer preferences and buying behaviors from different source systems around the multi-national company.

In mid-December, the world’s biggest beer company took a $50 million bet on a very different kind of beverage - a punt on pot, in fact.

Anheuser-Busch InBev (AB InBev), the maker of Budweiser, Stella Artois and Corona, announced a partnership with Canadian marijuana grower Tilray to research cannabis-infused, non-alcoholic drinks. Tilray will match AB InBev’s $50 million investment in the venture, which for now is limited to Canada, where recreational cannabis became legal in October 2018. According to Kyle Norrington, president of Labatt Breweries, a Canadian subsidiary of AB InBev:

We intend to develop a deeper understanding of non-alcohol beverages containing THC and CBD [chemicals produced by the cannabis plant] that will guide future decisions about potential commercial opportunities. We look forward to learning more about these beverages and this category in the months ahead.

In fact, AB InBev is somewhat late to the pot party. Drinks companies Constellation Brands, Molson Coors and Coca-Cola have already announced their own investments in cannabis. Either way, these deals are all part of a wider industry effort to deal with disruption in the beer market, where craft and local beers, and lighter, ‘healthier’ formulations are already having an impact.

Exploring taste, preferences and buying behavior

Naturally, there’s an important ‘big data’ aspect to better understanding the changes in customer taste and preference that drive beer-buying behaviors, says Harinder Singh, global director of data strategy and solution architecture at AB InBev. But until recently, he adds, getting that analytical work done was difficult for the company’s data scientists, business analysts and operational teams.

That’s because AB InBev’s strategy of growth through acquisition hasn’t just left it carrying a whole bunch of financial debt (currently in excess of $100 billion), but also shouldering technical debt, too, in the form of a data infrastructure littered with acquired tools and legacy technologies.

When he joined the company, Singh says he found 27 different ETL [extraction, transformation and load] tools, 20 different reporting tools and a huge number of disparate data silos, many hidden away in far-flung corners of the multinational company, feeding off well over 100 source systems. Singh says:

For me, this was a big technical challenge, clearly, but it was a big business challenge, too. Getting visibility into your customer, or your sales, or your supply chain, or your employees – all of those things become really difficult or take a very long time with a legacy architecture like this. In the past, that might have been fine, but as you move more and more towards being a digital business, every second, every minute that your team aren’t able to make business decisions, that translates to lost opportunities for the company as a whole.

Strategy and vision

On arrival at AB InBev, Harinder Singh quickly got to work, outlining a plan to build and deploy an ‘enterprise data hub’ -– a single, central repository where AB InBev’s critical data assets could be cleansed, standardized and integrated. As he recalls:

For me, 2017 was a year of communicating my strategy and vision - making sure that the company and its leadership understood the problem we were facing and making sure they understood that it wasn’t going to be fixed overnight, and then getting sign-off in terms of the strategy and budget to do what we needed to do.”

Then 2018 is mostly where I started on the practicalities: building a brand new data team in Brazil and Bangalore, because we really didn’t have much of a data engineering team to being with, and setting the goal of completing sufficient numbers of pilot projects to give the company’s leadership reason to believe that yes, this vision would be hard work and would cost money, but it was do-able.

The new enterprise data hub is based in the cloud on Microsoft Azure and uses Talend Data Preparation for data discovery and Talend Data Integration to bring together data from a wide range of sources - both from internal systems and from external, third-party data providers. The hub has a ‘golden layer’, where data resides once it is ready to be sliced and diced by internal users at the company. Talend also powers a more experimental ‘sandbox’ for data scientists, which they use to test various data models.

End-to-end visibility

While the enterprise hub has been designed to tackle the full range of data visibility challenges across the company, insight into sales and customers is a particular focus.

For example, AB InBev is already using it to analyse new trends among consumers for low-calorie beers and to identify how tastes for particular beers change, from season to season. This is based not just on its own internal data, but also consumer data from retailers and bars, from consumer research company Nielsen, from market surveys and from analysis of social media streams.

The company’s data scientists are also combining sales data with third-party data on weather and macroeconomics to predict what demand might look like at particular points in the future - offering big opportunities to optimize the company’s breweries and supply chain accordingly, says Singh.

His goal now for 2019, he says, is to have identified 100% of AB InBev’s critical data to the new infrastructure by the end of the year:

Yes, it’s an ambitious project, but when I look at how much work we got done in 2017 and 2018, I’m confident we can do it. It’s ambitious but it’s important as we move towards being a digital company that responds to customer needs and demands.