How grocery sensation Erewhon turned cultural buzz into business growth with dlt

Data stack

• Data sources: SQL DBs, MongoDB, APIs, clickstream data, Braze
• Destinations: BigQuery, GCS
• Orchestration: Dagster, Python
• Transformation: dbt

Challenge: Scaling data infra to meet rapid growth

Erewhon is an upscale, family-owned grocer and certified organic retailer in Los Angeles that curates ethically-sourced foods for customers committed to health & wellness. Despite having 10 markets and an online delivery service, its unmatched quality standards and unique inventory have driven global brand awareness through viral social media content. Erewhon is also a Certified B-Corp, using business as a force for good through its social and environmental performance, accountability, and transparency. The combination of quality and integrity has amplified Erewhon’s resonance with audiences worldwide.

Sean joined Erewhon as the first data hire, tasked with creating a straightforward analytics and reporting system using PowerBI. However, as the company's data needs expanded beyond simple reporting to support ambitious growth plans, it became clear that a more powerful and scalable solution was necessary. Sean, drawing from his experience as an analytics user of data platforms at Fortune 500 companies, recognized the value that advanced data systems could create. This insight became the driving force behind Erewhon's data transformation journey.

Erewhon’s rapid growth brought an increasing number of data sources and a greater need for data interoperation that exceeded the reporting and analytics features of PowerBI. For example, the data sources grew beyond legacy databases, adding new events from the Erewhon mobile ordering app and website, software APIs, integrated business ERP systems, and more.

"The secret sauce was choosing the right tools to decrease time to value and build efficiently. For me, some of those time savers are Dagster, dbt, and dlt. These tools allowed me to build enterprise-grade data operations at a fraction of the cost."

- Sean Pool, Data & Insights Manager, Erewhon

Fragmented data across the organization limited visibility into overall business performance and customer behavior. At the same time, reporting and analytics processes were manual efforts, which slowed down insights. Sean was in charge of driving the company forward with minimal developer resources available.

He needed to build a lean system to leverage data for business growth while facing the challenges of data quality, integration, access, and scalability. Solving each issue was essential for empowering the wider team to act upon more sophisticated use cases through data-driven app features, advanced marketing technology, personalized customer experiences, and streamlined process automation.

With existing infrastructure unable to support advanced goals, a transformation journey began.

Upskilling from BI to Data Engineering

With a background more aligned with business intelligence and finance than engineering, Sean was initially daunted by the task of creating a code-based data pipeline. “As a non-engineer, at first the idea of working with code was unapproachable and stressful,” Sean explains. “Getting started was a struggle, but I soon gained confidence with help from dlt's documentation, videos, and community support.”

Sean dedicated time to learning Python and advanced data engineering concepts, using resources like YouTube tutorials, AI coding assistants, and dlt's comprehensive documentation. This learning journey helped him solve the immediate data ingestion challenges and deepened his understanding of data engineering principles.

Solution: From low-code to high-code

“dlt simplifies everything about data ingestion.”

- Sean Pool, Data & Insights Manager, Erewhon

In order to build out a capable infrastructure, Erewhon’s existing setup needed to go beyond low-code solutions. This meant searching for alternatives that were accessible to those with light coding knowledge and powerful enough to handle complex problems. A lean team demands a lean implementation, but not all tools on the market allow for streamlined use or support those still learning advanced techniques.

Sean started off by creating pure Python pipelines before encountering dlt via YouTube. He evaluated other solutions in addition to dlt, but did not find other tools that were “nearly as clean and easy to implement, or robust.”

“I did look at Meltano and Airbyte, and Meltano was okay, but I thought dlt was better. I didn’t consider Fivetran given the costs there, and the pricing model is predatory in my opinion.”

- Sean Pool, Data & Insights Manager, Erewhon

Sean quickly developed his first pipeline with dlt, pulling data from MongoDB as a first source and, later developing connectors on top of the dlt REST API source.

Through a collection of resources, plus assistance from code companions and AI assistants like ChatGPT and Claude, Sean overcame various hurdles for more pipelines.

The dlt implementation is deployed on a cloud based VM, with Dagster handling orchestration and dbt handling data transformation. This created a fully automated, scalable pipeline from raw data ingestion to clean, analysis-ready data. Following software engineering best practices with CI/CD, the system also includes monitoring through Slack integrations that report on critical processes.

Sean explained that after "struggling through the specifics of our implementation," he realized how "simple and straightforward" it was to replicate the structure of his pipeline to add new data sources. Once established, he saw how much time was saved. Now, Erewhon uses dlt to implement data pipelines whenever possible.

“Dagster made it simple to do the orchestration of pure Python. Having one less step to configure it at a higher level with dlt made it so much easier to package.”

- Sean Pool, Data & Insights Manager, Erewhon

Results

The end result boasted a comprehensive data platform that uses dlt and streamlines multiple business functions. The new system integrates automated reporting, self-serve analytics, marketing automation, targeted advertising, mobile app support, systems integration, and AI workflows — all within a single platform.

Erewhon and Sean's journey showcases the potential of dlt for lean data teams and practitioners from non-technical backgrounds. By adopting dlt alongside complementary tools like Dagster and dbt, Sean built a powerful and efficient data platform that drives significant business value and innovation despite resource constraints.

This case study underscores the importance of modern data tools in democratizing data engineering and empowering lean teams to achieve ambitious data goals. dlt's intuitive design, Python-based approach, and reliability make it an ideal solution for organizations looking to enhance their data infrastructure without massive engineering teams or budgets.

Sean's success in building an enterprise-grade data platform as a solo practitioner highlights how the right tools can enable data-driven decision-making, regardless of team size or technical background.

Future: More dlt

Looking ahead, Erewhon plans to eventually replace existing custom pipelines using dlt. This will streamline operations, enhance reliability, and leverage dlt’s scalable framework for more efficient data management.

“I haven’t found any other tool that’s as clean or as robust for data ingestion. Once dlt gets data piping, you feel really secure, and you know it’s going to reliably work. Other tools and solutions had problems in that respect. Consider it as a serious option against any other tool, and differentiated in many respects, especially when scaling.”

- Sean Pool, Data & Insights Manager, Erewhon