Otrium drives ecommerce growth with personalization, saves €160,000

“Fivetran is the layer that brings all the data into our platform and lets us build a picture of the customer journey.”

- Rob Myers, Lead Data Engineer, Otrium

Key results

Increased the efficiency and reliability of code releases by a factor of 10
Enabled real-time data access for innovative projects including smart, algorithm-based pricing
Decreased pipeline maintenance issues by 87 percent
Freed up two data engineers’ worth of resources to focus on higher-value projects like data modeling

Otrium’s data stack

Pipelines: Fivetran
Destination: Databricks
Business Intelligence platform: Looker
Data sources: 30+ connectors, including PostgreSQL, MySQL, DynamoDB, Google Ads, Google Play, iTunes

Named one of the world’s most innovative fashion companies by Fast Company, Otrium is a trendsetting, global fashion platform with a mission to help reduce clothing waste. It partners with over 300 of the labels you know and love to offer older collections and excess inventory at affordable prices to more than 5 million customers.

Creating an accurate picture for engineers and analysts

As a startup, Otrium was hoping to improve its data platform to support the company’s explosive growth. Its lean data team was spending most of its time maintaining pipelines and supporting its legacy platform simply to keep it running. At the same time, the fragmented nature of the data and slow access to it was causing operational issues, including incorrect inventory numbers, and made it difficult for the marketing team to scale personalized campaigns and communications. Rob Myers, Lead Data Engineer at Otrium, explains:

“We were relatively resource light. We knew that we needed something that would get data, ingest it and drop it in tables. Once you decide you want to buy instead of build, you want it to be as easy as possible and Fivetran was the answer.”

Implementation of Fivetran was quick and simple, opening the data to engineers and analysts alike with an easy-to-use design. By leveraging the Fivetran platform with Databricks and a data validation tool, it’s now possible for anyone within the data teams to track where data comes from and create new tables.

“This implementation of Fivetran frees up time for our engineers previously spent on manual tasks, so they can focus on building much more robust and better data models,” adds Myers.

Building a Customer 360 and optimizing pricing

The core work of the data team is focused on the customer journey. From the moment a buyer logs into the website, through shipping and all the way to returns, everything is visible, logically stored and analyzed. The company is also syncing data from marketing tools and ecommerce tools to create the full picture of each customer.

“Fivetran is the layer that brings all the data into our platform and lets us build a picture of the customer journey.”

-Rob Myers, Lead Data Engineer at Otrium

This comprehensive, 360-degree view of consumer behavior was just the beginning. Otrium’s new-found ability to gather and centralize data also supports organizational efforts. This includes predicting product returns and enabling smart pricing, wherein the company scrapes the prices of competitors for similar products across the market. Based on this data, an algorithm optimizes Otrium’s prices several times per week — keeping them competitive and directly impacting revenue.

Enabling efficient data engineering and innovation

With all the data improvements, awareness and democratization now present at Otrium, the biggest win for Myers and his team is saving €160,000 annually on data engineering costs, with the added bonus of freeing up the team to focus on high-priority projects thanks to over 30 Fivetran connectors.

The engineering team’s efficiency has also improved dramatically with over 30 code releases per month. This is a tenfold increase from the previous two to three monthly releases, which were mainly fixes to support their legacy solution.

The reliability of code releases has also improved. Where before, according to Myers, code releases were “like handling an expensive fragile vase,” there is increased trust in them now and they can quickly adjust in case of problems. What’s more, the reduction in pipeline issues has been dramatic, from two per week down to one per month.

In a very short time, Otrium has built a robust data infrastructure that can serve as the foundation for its continuing expansion. More work is underway, including predicting and addressing customer behavior using machine learning and smarter algorithms around product recommendations and customer communications.

“It’s still early days, but we’re looking at product recommendations and improved customer communications,” comments Myers. “Having a complete picture of our data and a full view of consumers is really important to innovative projects. It requires the level and breadth of data that we now have, and Fivetran got us to the place where we can start doing that really interesting stuff.”