A data engineer builds and maintains the systems that move data from the places it is created into one place where it can be trusted. Day to day, that means connecting new data sources, writing the transformations that turn raw tables into clean ones, monitoring pipelines and fixing them when they break, checking data quality, and controlling who can see what. In an ecommerce brand, the data engineer is the reason that revenue in the finance report, the marketing dashboard and the board deck is the same number. Brand Media Works provides this work to ecommerce brands, so the examples below come from that world.
What does a data engineer do day to day?
- Connects sources. Sets up and configures the pipelines that pull data from Shopify, Meta Ads, Google Ads, Klaviyo, the support desk and finance systems, usually with an ingestion tool or custom code.
- Builds the warehouse. Designs where each source lands, how history is kept, and how tables are organized so they can be joined.
- Writes transformations. Turns raw tables into agreed definitions, such as net revenue, new versus returning customers and blended acquisition cost, usually in SQL with a modeling tool such as dbt.
- Monitors and fixes pipelines. Watches for failed syncs, missing days and duplicated records, and repairs them before anyone reports on bad numbers.
- Manages access and security. Decides who can see customer-level data, masks personal fields where they are not needed, and keeps credentials and tokens current.
- Serves data to other tools. Makes clean tables available to dashboards, sends audience lists back to ad and email platforms, and feeds AI tools.
Why is so much of the job maintenance?
Because the sources keep changing. A pipeline that worked last quarter can break or quietly return the wrong numbers when a platform changes its API. Recent changes that every ecommerce data engineer has had to handle include:
- Shopify deprecated its REST Admin API on 1 October 2024, and began moving shops' shipping settings into Markets from 1 October 2026, after which the old shipping data is a frozen snapshot.
- Meta removed the 7-day and 28-day view-through attribution windows from its Ads Insights API on 12 January 2026, which changes how conversions are reported.
- Google stopped issuing new Google Ads developer tokens on 9 September 2026.
- Meta requires an app to make at least 500 Marketing API calls in 15 days, with an error rate under 15%, to keep full API access.
On top of platform changes, numbers change after the fact. Klaviyo keeps revising conversion attribution for up to 5 days after a send by default, and up to 90 days if the window has been raised, so the pipeline has to go back and re-read recent data rather than copy each day once.
How is a data engineer different from a data analyst?
| Role | Main job | Typical tools |
|---|---|---|
| Data engineer | Makes data complete, correct and available in one place | Ingestion tools, a cloud warehouse, SQL, Python, orchestration |
| Analytics engineer | Turns warehouse tables into agreed business definitions | SQL, dbt, the warehouse |
| Data analyst | Explains what happened and why | Dashboards, spreadsheets, SQL |
| Data scientist | Forecasts, models and experiments | Python, statistics, machine learning tools |
In a small team one person may cover two of these. The risk is when an analyst ends up doing data engineering by hand, exporting and reconciling spreadsheets each week, because that work never becomes a pipeline anyone else can rely on.
What skills does a data engineer need?
SQL above all, because most transformation work is written in it. Python for custom connectors and automation. Working knowledge of at least one cloud warehouse, such as BigQuery, Snowflake or Redshift, and of an ingestion tool. Data modeling, so tables join cleanly and definitions stay consistent. And an understanding of the business, because deciding how revenue treats refunds or how a returning customer is defined is a business decision the engineer has to build correctly.
Does an ecommerce brand need its own data engineer?
Not at the start. Published US rates for a data engineer run from $70 to $150 an hour, so a month of one engineer's time comes to $11,200 to $24,000 (our arithmetic, at 160 hours). A first pipeline from one or two sources is published at about $4,000 and two weeks as a fixed-price project, which is a cheaper way to learn what the data can do.
A full-time hire makes sense once the data work is continuous: several sources, a warehouse, regular new requests from finance and marketing, and AI tools that depend on the data. Until then, a partner that builds the first pipelines and handles maintenance part-time covers the job. How much marketing data engineering costs sets out published project prices, and Data engineering: a guide for ecommerce brands covers the sources, the stack and where to start. Marketing data engineering describes how Brand Media Works does this work.
Frequently asked questions
Is a data engineer the same as a data scientist?
No. A data engineer builds and runs the pipelines and tables that data scientists and analysts use. A data scientist builds forecasts and models on top of that data.
Can a marketing team do data engineering with no-code tools?
Partly. Ingestion tools can connect common sources without code. The harder parts, agreeing definitions, keeping history, and catching broken or revised data, still need someone who owns the pipeline.
What does a data engineer do in the first month at an ecommerce brand?
Typically: list every source and who owns it, connect the two or three that answer the most urgent question, load the history the platforms still hold, and agree definitions of revenue and acquisition cost with finance and marketing.