Fivetran

Schüttflix, a German logistics start-up, aimed to disrupt the traditional construction supply chain by digitizing the industry, which was primarily reliant on pen and paper processes. The company's mission was to enable data-driven decision-making by providing stakeholders with accurate and timely data. The challenge was to build a modern data stack that could tap into key data sources efficiently and reliably. Alexander Rupp, Head of Data and Business Intelligence at Schüttflix, was tasked with evaluating connectors that could meet these requirements.

Lendi, an Australian mortgage broker with over $12 billion AUD in home loan settlements, was facing a significant challenge in its data management. The company's business model, which involves offering customers a choice of over 2,000 loan products from more than 40 lenders, relies heavily on delivering the right experience to the right person at the right time on the right digital platform. This requires accurate insights into borrowers' needs and preferences, which in turn requires tapping into behavioral data on third-party engagement platforms such as Facebook, Google, and Bing. However, the data from these platforms were siloed and did not integrate easily. Even when the data could be brought into the same repository, the data structure was often inconsistent, necessitating data cleaning before it could be used. The responsibility of ensuring that stakeholders across the company could analyze and create actionable insights from this third-party data fell to Lendi’s Data Architect, Daniel Deng.

Paytronix, a customer engagement platform for restaurants and small businesses, was facing a significant challenge in managing and deriving insights from its data. The company was dealing with data from multiple sources, running on various databases, and in disparate formats. The data ingestion tool they were using was unreliable and missed many transactions, leading to a lack of trust in the underlying data. Additionally, the company was using a mix of Scala and PySpark jobs for data transformation, which was custom code and handwritten. This toolset was unable to keep up with the growing demands of the business, and a lot of time was spent on maintenance and break-fix support. The company wanted to focus more on experimentation, but the existing system was not conducive to quick proof-of-concept testing and rapid iteration.