AB InBev moves to Talend to streamline IT estate and free up data science talent


AB InBev has grown from a medieval abbey brewery in southern Belgium into the world's largest beer maker through a series of mergers and acquisitions that left it with a sprawling estate of IT systems spanning 27 different ERP systems and more than 20 Extract, Transform, and Load (ETL) tools.

To integrate all its data into a unified view and enhance its analytics capabilities, AB InBev replaced all these legacy ETL tools with the Talend data integration platform.

All the data the company collects from ERP systems, social media, sensors in breweries, Nielsen research is pooled into a central depository in an enterprise data hub that lives on the Microsoft Azure cloud. The data is then analysed to understand consumers, products and every aspect of the business, from changing tastes in drinks to optimising supply chains.

The move to Talend has freed up the scarce data science talent that AB InBev depends on to discover crucial business insights.

When the data was in silos, they had to find the specific system used by each department in every regional area and then analyse the information in comparison with what they found in the other systems. This meant that they spent 80 percent of their time on just finding the data, leaving only 20 percent of their capacity free for analysis.

"Now they can spend most of their time actually analysing the data because they already have access to it," says Harinder Singh, Global Director of Data Strategy and Solution at AB InBev.  "They have a data catalogue to go with that and they can rely on the documentation that goes with it to understand where the data is coming from and what it means."

Improving decisions for different markets

Talend has helped AB InBev reduce time to market in the more than 100 countries where its 500 beers are served, but it has also helped with the tricky task of making global decisions.

"In most of those cases that used to take let's say three months," says Singh. "Now that could be done in three weeks or three days, because we have already built the pipeline in a reusable way."

The increased speed has had a major impact on demand forecasting.

These predictions vary immensely across different AB InBev markets depending on a vast array of seasonal influences, from local economics to festivals and weather.

Weather forecast influences the crops grown at local farms, their transportation in trucks to the brewery, the production process there and the strategy for distributing the finished beers to bars and shops.

To understand the impact of all these factors, AB InBev created a demand forecasting model based on sales data from all the different regions.

"We sell globally, but we brew and produce locally, so all of those pieces of data we have to pull into our demand forecasting," says Singh. "It helps our marketing team and sales team to understand and our operations in their planning processes."

Read next: How TfL is using predictive analytics to keep the Underground moving

They also use the data integrated in Talend to predict future failures of machines fitted with sensors and to track the temperature of beers through RFID codes attached to kegs.

Once the beers hit the market, they can quickly analyse consumption habits to guide the next season's production strategy.

This gave AB InBev the time to prepare for the growing demand for light beer brands, which now make up the three best-selling beers in the US.

"In developed countries, we noticed that people are interested in trying new flavours or brands, and people are in some cases making healthier choices," says Singh. "All of that insight feeds into our new product development."

Why Talend?

AB InBev turned to Talend after analysing all the tools that the company had accumulated through mergers and acquisitions as well as new players on the market. Each individual tool was rated in a variety of areas including costs, features, user-friendliness and product roadmap, which were then added up into a total score.

Singh feels that Talend's future capabilities were perhaps its greatest strength.

"That tells us where the company will be going forward," he says. "Everything is changing very quickly so we wanted to make sure we worked with a company that can change at our pace."

Talend further stood out from the competition by offering a reusable data extraction framework for all of ABInBev's markets, connections to all of the systems in the company's vast IT estate and compatibility with all leading cloud vendors.

It also provided value as it charged a price per user rather than per data connection.

Read next: What is the UK government's open data strategy?

Setting up Talend in ABInBev's Azure cloud, however, was not a simple process.

"It's a product that's very capable but it does need specialists to install and configure it,” says Singh. "The way we did that is we partnered with Talend's professional services so we could make sure that we were getting the right implementation and configuration, because in the foundation any changes you make at that level could have a long-lasting effect, and we wanted to make sure that we pulled in the specialists to do it right."

To those who are nearer to the beginning of their analytics journey, Singh warns that no single solution is complete.

"Every product has its strengths and weaknesses, so the best way to go about is to first understand what the business needs are," he says. "Number two, check if the product works well with the rest of the ecosystem, because every company already has applications.

"Three, make sure that you have the skill sets and the resources for using all these different dimensions when making the decision. Don't just look at one aspect or not just look at cost. Obviously, cost is always important but there are many more factors that should be included in the decision-making process."

Copyright © 2018 IDG Communications, Inc.

Download: EMM vendor comparison chart 2019