A cloud-based brand management platform company engineers a trusted data Ecosystem with the Snowflake Data Cloud
San Francisco, California
Data Sharing, Data Science, Data Engineering, Snaowflake Data Cloud
The company’s vision is to create a world where everyone is a part of building beloved brands. Its Saas platform brings everyone and everything brand related together under one ecosystem, including brand guidelines, DAM, a space for collaboration, and more. A unique network of integrations, an open developer toolkit, and upcoming offerings for expanding the ecosystem ensure brands at every scale can evolve and thrive with them.
Maintaining query performance and trust as data volumes grew, The company has undergone a rapid journey from startup to global brand management leader. But as it grew, so did its data. “When I first joined the company, we were a team of 80 people,” Head of Data, Sibel Atasoy Wuersch, explained. “But even for a business of that size, we generated a lot of data across all our tools.”
At first, the company used custom Python scripts to move data into a MySQL database and then used Tableau for reporting. But over time, its database performance fell short of expectations.
“We were seeing issues with our scripts failing, resulting in inconsistent, unreliable data,” Atasoy Wuersch said. “It took too long for our leadership team to get the insights they needed, and it wasn’t scalable.”
A proven, modern, and future-proof data platform
Atasoy Wuersch and her team began comparing data platform solutions. And as they ran proof-of-concept (PoC) projects, the choice quickly became clear.
“We considered a wide set of criteria,” Atasoy Wuersch said. “How easy is it to connect to different sources? How easy is it to manage access? What about costs? Is it flexible? When we asked these questions, one platform had all the answers: Snowflake was the clear winner.”
Over three months, they completed two proof-of-concept projects, including one comparing Fivetran and Stitch. It ultimately decided to use Snowflake and Fivetran and spent nine months migrating workloads and connecting larger data sets through custom scripts.
We [moved] quickly, learned from each stage, and ultimately solved some of our most pressing challenges around data validation and trust.
MICHAL LAPINSKI, Data Infrastructure Lead
A Simple setup helps them to experiment and build a robust data pipeline
Thanks to Snowflake’s simple documentation, the team managed its two PoC projects simultaneously with just one data engineer, Lukas Jaeger. “The switch from our MySQL environment to Snowflake was quite straightforward and we only needed to change the destination database, and we were basically set up,” Jaeger said. They implemented Snowflake across several phases. At first, it integrated data sources into Snowflake through Fivetran, but this phase only included ingestion and storing of raw data. Phase two involved designing the database and developing data marts to analyze further data stored in Snowflake. They also brought in ThoughtSpot during this phase—so the team had to get its data marts up and running while simultaneously preparing data for initial ThoughtSpot use cases.
They also restructured their data team around this time. It brought in Michal Lapinski, Data Infrastructure Lead to ensure the final deployment stages went smoothly. And once they had launched ThoughtSpot, it expanded the team further to scale data discovery, validation, and delivery and build awareness, adoption, and enablement.
“There were many things we carefully planned and refreshed multiple times, like our data schemas and databases, and approach to user access rights,” Lapinski recalled. “We took a few detours, but we took them quickly, learned from each stage, and ultimately solved some of our most pressing challenges around data validation and trust.”
Trusted data powers, trusted business decisions.
ThoughtSpot and Snowflake deliver accurate, trusted information that helps their leaders and employees confidently make decisions. “Sales, marketing, customer success, finance, and even our product teams are all using Snowflake and ThoughtSpot to access data,” Lapinski explained. “ThoughtSpot Liveboards, powered by our data pipeline in Snowflake, help these teams analyze performance against our KPIs every week to make sure the business is on track to meet our goals.
And if they aren’t quite meeting those goals, teams trust they have the accurate data to identify the problem and find a solution. “Having a central source of truth to show where people are going wrong is really valuable,” Lapinski said. “When we connect all our tools through Snowflake and ThoughtSpot, we can see when we have gaps.
“What we’ve achieved in ThoughtSpot needs a powerful environment where the data is designed properly,” Lapinski explained. “We do that through Snowflake. We use the platform to engineer pipelines and develop the backbone to power self-service analytics for our users.” The figures speak for themselves. The teams see average query response times of 0.22 seconds across 660 million rows of data in ThoughtSpot production environments. “All I can say is thank you, Snowflake, for the fantastic performance,” Lapinski said. “We simply couldn’t have done this before.
Thank you, Snowflake, for the fantastic performance. We simply couldn’t have done this before.
MICHAL LAPINSKI, Data Infrastructure Lead