PayU GPO Creates a Unified Global Data-Driven Culture with Snowflake’s AI Data Cloud

Since its founding in 2002, Netherlands-based PayU GPO has grown into a leading global online payments service provider, operating in more than 50 markets across Latin America, Central and Eastern Europe and Africa. 

Having expanded largely through acquisitions, PayU GPO functions under a federated business model, with various subsidiaries operating independently under a unified banner. This enables PayU to maintain local expertise — at a global scale — but it also means its analytics landscape was fractured, made up of various data platforms that prevented PayU GPO from gaining holistic insights into its operations. 

“Data is a critical component in our business,” says Piotr Wojda, Head of Analytical Data Platform at PayU GPO. “These various platforms had different processing methods, and there was no common language. We were looking for a tool that would allow us to combine this data at the global level without the challenges related to those individual infrastructures.”

To meet this need, PayU GPO implemented Snowflake’s AI Data Cloud at the core of its data activities in 2020.

The first step on PayU GPO’s journey to create a data-driven culture across the company was getting all of its data in one place. The company built 800 pipelines to funnel data into Snowflake, making it accessible to anyone who needed it. But soon, it became apparent that there was a significant obstacle to overcome.

“Our various platform teams all used entirely different terms and measurement techniques,” Wojda says. “To promote global analytics and compare apples to apples, we needed to ensure we had a unified terminology.”

With all of its data now in Snowflake, the company’s data team could begin to understand and make sense of the confusion caused by this internal language barrier. PayU GPO then created a unified glossary of terms — a vital first step on the way to global understanding.

“From there, we started to build global reports, global data models and global dashboards that were certified, so everyone started seeing the same numbers and understanding them in the same way,” Wojda says. 

Importantly, PayU GPO was able to do this without disrupting the internal cultures and practices at each of the various subsidiaries. Snowflake’s wide range of integrations and compatibility allow data specialists to continue using their preferred tools and languages, whether it be dbt,Tableau, Airflow, REST API’s, Java or Python. There is no need for onboarding or retraining, and data has quickly and seamlessly become democratized across the organization.

“We now have about 350 active Snowflake users, which is around a third of the company,” says Wojda. “They’re using Snowflake on a day-to-day basis, from business and operations teams to marketing and sales.”

Five years on from the start of this data transformation, PayU GPO has been able to reflect on the capabilities the AI Data Cloud has provided — and what they have helped the company achieve.

Thanks to Snowflake’s scalable cloud-based infrastructure and separation of storage and compute, PayU GPO has been able to dramatically reduce the time it takes to run models and implemented further optimizations based on recommendations from Snowflake architects. This led to a 36% initial reduction in TCO, and daily model calculations that previously took almost three and a half hours to complete now take only 40 minutes. 

The company has also been able to ingest data 83% faster (from 2 hours and 30 minutes, down to just 50 minutes), update financial models 73% faster and reduce overall data refresh times from 24 hours to just 15 minutes.

Snowflake’s ease of use and PayU GPO’s efforts to democratize data led to users in every corner of PayU GPO’s business relying on Snowflake and the insights it holds. But not all users were adhering to the most up-to-date best practices, and as a result, there were inefficiencies in resource utilization.

To address this problem, PayU GPO sought advice from Snowflake Professional Services. 

“As data people, we don’t solve problems on hunches. So the first thing we did was use Snowflake itself to identify which areas were the most inefficient," Wojda says. “The Professional Services team was then extremely helpful in providing us with various techniques and strategies to enhance our efficiency.”

Armed with these insights and guidance, PayU GPO reduced the run frequency and refreshes of many of its data models. Optimizations to processing logic were also applied, for example by using Snowflake Dynamic Tables, which made a significant impact on efficiency. In fact, the costs associated with individual models were, in some cases, reduced by up to 94%, from $850 a month to only $50 a month. 

Overall, PayU GPO was able to save 20% in costs across its Snowflake deployment, thanks to the Professional Services team.

PayU GPO plans to continue working with Snowflake’s experts to explore new features and implement further optimizations across its data stack, facilitated by a monthly catchup between the two parties. 

“These meetings are really helpful,” Wojda says. “We can use our account team to learn about new features and security recommendations, and to troubleshoot any issues we have.”

Of particular interest right now is Cortex AI, which could help support PayU GPO’s emerging AI initiatives. But the company also plans to introduce data mesh principals, improve governance and implement more data sharing — both within the company and with external partners. 

“We really appreciate that Snowflake is adding more and more features that we can quickly start using in our day-to-day work,” Wojda says. “Snowflake has been a great enabler for us to expand the way we work with data, become more data-driven and get more value out of the data we have.”