Upsell excels in brand marketing across all channels by creating and managing dedicated, high-performance, long-term sales teams for each client.https://www.upsell.fr/
Find out how Upsell's lean data team modernized their legacy data stack and made over 5000 undocumented assets discoverable.
Upsell is one of the fastest growing forward deployed sales organizations in France. They help their clients build and grow their brand across a rich mix of distribution channels.
Tidiane Ndir, Chief Data Officer, joined the company in March 2023, and found himself inheriting a stack consisting of an in-house ETL tool, warehouse, and no documentation to provide any context on over 5,000 undocumented data resources.
Data democratization and reducing the cost of onboarding new data team members was identified as a priority. Using Secoda’s automated data catalog and lineage, Tidiane and his team migrated Upsell’s legacy infrastructure to a modern data stack. In parallel, Tidiane used Secoda AI to generate documentation and accelerate time to value for new data team hires. With a data team of 3, Tidiane has been able to create over $400K in annual cost savings by improving self-serve data discovery and reducing time to value.
With no official documentation for their data assets, the same questions would be repeatedly asked by business users. It was challenging to efficiently answer basic questions about Upsell’s data and increase adoption of self-serve analytics.
With the company growing, the data team also needed to scale, and bringing on new data team members meant starting from scratch, with each new hire facing a steep learning curve requiring over two weeks to become autonomous with data assets. Tidiane wanted to find a tool that would make documentation easy, and be compatible with the new, modern stack he was migrating to (Panoply, Bigquery, and Metabase).
"Before Secoda AI, we were struggling to answer even basic data questions. We had to rely on our data analysts to do everything, which was time-consuming and expensive"
Upsell recognized that the solution to their data problems lay in implementing a comprehensive data management platform including AI powered documentation, data cataloging, lineage and monitoring - Secoda.
Tidiane wanted a tool that would be able to automatically bring in any new data assets in the stack, be user-friendly and have an easy to use interface, and give him the ability to write documentation as quickly and efficiently as possible.
Tidiane and his team leverage Secoda’s AI assistant frequently. They established a new principle for the team, that every data asset and process must be documented. With a data team of 3, and 5,000 undocumented resources - this was a huge challenge. Using Secoda’s AI assistant to generate documentation, the team was able to reduce time to documenting data assets by 90%.
Upsell is based in France, and the team was also able to leverage Secoda AI to document data team processes in plain French, so that anyone new on the team can understand how the data team functions. Where it previously took 2 weeks to get analysts up to speed on the team, with Secoda, the team was able to get new analysts working autonomously with data assets in under 3 days.
“Secoda AI has given us the power to answer any data question ourselves, regardless of complexity. This has saved us a lot of time and money, and it has empowered our business users to be more data-driven. The documentation has helped our business users and new data team members be autonomous.”
With Secoda AI’s SQL generator, Upsell is able to generate SQL queries in French, which is extremely useful for the data team, as well as business users who need some extra assistance.
Since adopting Secoda, Upsell's data team has seen transformation. With time spent on documentation and onboarding dramatically reduced, the data team is able to focus on their main priorities, enhancing productivity and saving valuable time.
“We have absolutely accomplished much more than we initially expected. We’re a data team of 3 looking after many large customers, and if we didn’t find the right tools for our use cases, we would have struggled to satisfy our clients. With this new, optimized stack, we expect to continue to scale effectively.”