What Does an Analytics Engineer Actually Do?
An analytics engineer is the person in the data team who owns the transformation layer — the clean, tested SQL models between raw ingested data and the dashboards analysts use. This guide covers what the job involves, how it differs from data engineering and analysis, and whether it’s right for you.
- What an analytics engineer is
- Analytics engineer vs data engineer vs data analyst
- What an analytics engineer does day to day
- The skills and tools you need
- A day in the life: tracing a broken revenue number
- How to become an analytics engineer
- Is analytics engineering a good career?
- Frequently asked questions
What an analytics engineer is
Before dbt and cloud warehouses, transformation hid in one of two places: a data engineer’s pipeline code, where analysts rarely ventured, or BI-tool logic and spreadsheets, where it drifted quietly for months. Those tools made it practical for analysts to write SQL like software engineers — version-controlled, peer-reviewed, and tested before it ships.
Analytics engineering filled that gap. The role owns the transformation layer: dbt models, tests, column-level documentation, and often a semantic or metrics layer that defines business terms — what “revenue” means, how “active user” is counted — so every dashboard downstream reads from the same definition.
The gradient matters: an analytics engineer does less heavy infrastructure work than a data engineer (no cluster management, lighter Python) and less statistical work than a data scientist, but more engineering rigor than a typical analyst.
Analytics engineer vs data engineer vs data analyst
The three roles form a chain. Data engineers get data into the warehouse; analytics engineers shape it into trusted models; analysts ask questions of those models. Here is where each sits:
| Role | Owns | Main tools | Typical output |
|---|---|---|---|
| Data Engineer | Pipelines & infrastructure | Python, Airflow, Spark, cloud platforms | Data reliably landed in the warehouse |
| Analytics Engineer | Transformation layer | dbt, SQL, git, warehouse | Clean models, tests, metric definitions |
| Data Analyst | Answers & dashboards | SQL, BI tool, spreadsheets | Reports, charts, ad-hoc analysis |
The boundary between analytics engineer and data engineer blurs at small companies, where a two-person team often shares both roles; at larger teams the tools and daily focus split them cleanly.
What an analytics engineer does day to day
A typical week involves most of these:
- Writing and refactoring dbt models — SQL SELECT statements organized into staging, intermediate, and mart layers that run inside the warehouse on a schedule.
- Maintaining tests — not-null checks, uniqueness assertions, referential integrity — automated checks that catch a broken model before an analyst spots a wrong number in a chart.
- Writing documentation — column-level descriptions so “net_revenue” means the same thing to engineering, finance, and marketing.
- Maintaining the semantic or metrics layer — centralized business-metric definitions that every dashboard reads from, so each team isn’t calculating its own version of the same number.
- Debugging discrepancies — tracing a number that looks wrong back through models and sources to find the root cause.
The testing and documentation work is underrated — production-grade SQL without tests is just a query.
The skills and tools you need
You don’t need to be a software engineer, but you do need to work like one in a few areas:
- SQL — advanced fluency: window functions, CTEs, complex joins, performance tuning. This is the foundation everything else builds on.
- dbt: the dominant tool for analytics engineering — models, tests, docs, snapshots, Jinja macros. See dbt for beginners to get started.
- Git: version control for SQL is the habit that most separates analytics engineering from ad-hoc querying — branch, pull request, review, merge.
- Data modeling: star schema, fact and dimension tables, dimensional modeling patterns — the structures that make warehouse data query-fast and metric-ready.
- Light Python: Jinja macros in dbt, data quality scripts, occasional automation. Not Spark jobs, not infrastructure management — that’s the data engineer’s lane.
A day in the life: tracing a broken revenue number
A Slack message lands Monday morning: “Revenue shows $40k last week; Finance says $47k. Any idea what’s off?”
Here’s how an analytics engineer traces it:
- Open the dbt model
fct_revenueand read the SQL: it joins orders to refunds. - Run dbt tests in a dev environment. A uniqueness test on
order_idfails: a one-to-many join has fanned out orders that carry multiple partial refunds, over-subtracting before the totals aggregate. - Fix: add
DISTINCTto the orders CTE. Add a uniqueness test on the final grain. - Open a pull request. CI runs dbt; all tests pass. Merge and reply: “Revenue model patched — fan-out from the refunds join. Uniqueness test added.”
No pipeline change from the data engineer, no fix from the analyst — the analytics engineer owned the model in between, and left it more robust.
How to become an analytics engineer
Two paths are most common.
Path 1 — analyst who picks up engineering habits. You already understand the business questions, the data, and the BI tool. The step into analytics engineering is learning git, dbt, and pull-request workflow — habits that take weeks to build, not years. Most analytics engineers arrive from this direction.
Path 2 — data engineer moving up-stack. You have Python and pipeline experience; the move is developing stakeholder communication, business-metric fluency, and SQL modeling patterns. It fits if you like infrastructure but want closer contact with business questions. See how to become a data engineer for a direct comparison of where the two roles diverge.
Either path benefits from learning dbt properly and from understanding dimensional modeling — star schema explained walks through the core vocabulary.
Is analytics engineering a good career?
Demand tracks adoption. Companies on dbt and cloud warehouses typically have a distinct analytics-engineering function; those still on legacy on-premise stacks may not carry the title at all. Check whether target employers are inside the dbt ecosystem before tailoring your resume to this role.
Compensation varies by market, seniority, and company type. Use current salary surveys for your region — published figures age quickly.
The fit is real, and the role is still settling. For people who love SQL and data modeling but don’t want to manage infrastructure, analytics engineering is a satisfying seat — engineering rigor applied to business logic, with direct stakeholder impact.
Frequently asked questions
Analytics engineer vs data engineer — what’s the real difference?
Data engineers build the pipelines that land data in the warehouse — orchestration, connectors, infrastructure. Analytics engineers work inside the warehouse: they model, test, and document the data once it arrives. A DE asks “did the data arrive?”; an AE asks “does it mean what we think it means?” The line blurs at small companies, but the tools and focus are distinct.
Do analytics engineers need Python?
Less than a data engineer. Most AE work is SQL and dbt, which uses Jinja (not Python) for templating. Python shows up in data quality scripts, macros, or lightweight automations. Writing Spark jobs or managing cloud infrastructure is the data engineer’s territory.
Is analytics engineering a junior or senior role?
Both. Junior analytics engineers often come from analyst backgrounds and learn git and dbt on the job. Senior roles involve architectural decisions: how the semantic layer is organized, which metrics get centralized, when to push back on a modeling request.
What tools does an analytics engineer typically use?
The core stack: dbt for transformation, a cloud data warehouse (Snowflake, BigQuery, Databricks, and others), git for version control, and a BI tool downstream. The modern data stack overview shows how they fit together.
The analytics engineer is the link that makes warehouse data trustworthy at scale. To see where the role fits, start with the modern data stack overview; and if you’re weighing this path against the data engineering track, how to become a data engineer maps out where the two roles diverge.
Last updated: July 6, 2026
Comments
Post a Comment