Improving affiliate marketing ROI with customer insights

Uniting data operations to meet new digital demand

Awin Global provides an affiliate marketing network that helps connect people with the products and services they demand while bringing greater visibility and increased revenue to companies online.

With such a vast network of advertisers and their customers, Awin has access to massive amounts of highly valuable data, including engagement and transaction details across partners and customers. Publishers and partners depend on Awin’s recommendations being both current and accurate, using the insights gleaned to make business decisions about campaign budgets, new product offerings, discounts and marketing strategies.

Over time, this data has become exponentially more useful, serving as a critical piece of how the business functions and innovates. However, this data has also become significantly more challenging to manage and leverage as the company has grown.

“We ingest, analyze and export a terabyte of data every single day. From an engineering perspective, we were not able to leverage this data and deliver insights in a timely manner for internal and external stakeholders to make smarter decisions,” said Anup Belbase, Awin’s Head of Data Warehousing.

Despite a dedicated approach to data management, the sheer volume of transactional and other data clogged up the pipeline, resulting in day-long waits for insights that delayed projects and outcomes for data-reliant teams throughout the organization. Awin needed to revisit its data analytics infrastructure to unlock the potential of all its data, and they chose Databricks to make it happen.

Democratizing data, ML and BI for smarter campaigns

With a future goal of creating a robust center of excellence (CoE) to democratize the data analytics process, Awin is now leveraging the unmatched speed and computing power of the Databricks Data Intelligence Platform to generate more explicit, effective and accurate assessments for the brands they work with. Now they are able to consistently provide new insights that enable those businesses to turn innovative ideas and consumer behavior into actionable programs that convert at a higher rate. By integrating BI tools like AAS and Tableau, Awin’s team can easily and quickly surface key information across hundreds of reports and dashboards — and analyze it with pinpoint precision — in far less time, and with fewer resources, than before.

With Delta Lake, Awin now has a common data layer that federates all their data to support all business decisions and ML. And as data continues to flow daily, version control and ACID compliance ensure consistency, accuracy and peace of mind. “Delta Lake gives us the confidence that we are making decisions based on complete and accurate data,” said Belbase. “And if anything goes wrong, we can always roll back to a previous version of the data. Our old system never would have allowed us to do this.”

Upping the actionable value of big data

Today, the data ingestion-to-insights process, which once required a full day, now only takes around an hour or less. With data at their fingertips when they need it, Awin has leveraged ML for process automation now and a recommendation engine that identifies beneficial partnerships and opportunities between advertisers and publishers that yield the highest success rates.

“The impact of our recommendation engine has been significant for our customers,” said Ben Poole, VP of Business Technology at Awin Global. “With improved insights, we are able to recommend publishers to our customers that give them the best chance of driving conversion. With ML-powered recommendations, we’ve seen a 9x improvement in the rate our customers accept our publisher recommendations. This engine is in its early stages, but will drive revenue growth through next year and beyond.”

Databricks has helped Awin significantly lower the cost of business operations, as well as reduce ETL windows by over 40%. This has subsequently resulted in 20% cost savings on processing and cutting operational IT costs; moving to a service-based architecture also reduced hosting fees by about $25K per month. Productivity has vastly improved, allowing for faster time-to-market, more predictable outcomes and more successes for their clients.

“Our efficiency and accuracy have improved, and people depend on that speedy access to these insights to move the business forward,” explained Poole.

Over the next 18 months, Awin will continue to leverage the Databricks Data Intelligence Platform to expand their ML portfolio into churn prediction, demand forecasting and anomaly detection for fraud prevention — all with the ultimate goal of improving business operations and driving revenue for their customers.

“Databricks has been instrumental in allowing us to successfully complete our data analytics and machine learning journey by providing both guidance and expertise to solve our toughest data problems,” concluded Poole.