Analytics engineering is the role that didn’t exist in 2018 and has become the natural second job for a lot of analysts by 2026. It sits in the gap between "classic analyst writing ad-hoc queries" and "data engineer building pipelines," owned by people who know enough SQL to be dangerous and enough software-engineering discipline to ship production models. This post is the career-change read for analysts wondering whether the jump makes sense.
What an analytics engineer actually does
The job title describes the tool stack: SQL, version control, dbt or a dbt-alike, a warehouse (Snowflake / BigQuery / Redshift), a BI layer on top. The actual day-to-day is:
- Model building — translating raw-source tables into clean, tested, documented "mart" tables that analysts and the BI layer consume. The bulk of the job.
- Test writing — schema tests (uniqueness, not-null, referential integrity), plus data tests (this column should never be negative, this metric should not drop 20% WoW without a reason).
- Code review — every change goes through a PR. Reviewing others’ models, defending your own.
- Documentation — dbt docs, metric definitions, lineage. If the analyst team doesn’t know what
fct_purchases_cleanmeans, you’ve failed. - Stakeholder translation — less than in a pure analyst role, but still real. The analysts ARE your stakeholders, plus whichever product/finance team owns the source data.
Who should make this jump (and who shouldn’t)
Analytics engineering is a great next step for analysts who:
- Like writing SQL more than presenting findings. The role is 70% SQL, 15% review/writing, 15% meetings. Analyst roles often flip that ratio.
- Get frustrated by inconsistent metrics. If you’ve ever been driven crazy by "revenue" meaning three different things across three dashboards, analytics engineering is the role that fixes that problem.
- Enjoy engineering discipline. Version control, code review, testing, CI. If that appeals to you, you’ll be happy. If it sounds like bureaucracy, you’ll be miserable.
It’s a worse fit for analysts who love stakeholder work, live for the "aha" moment of presenting a finding, or get their energy from open-ended exploration. That’s not engineering; that’s classic analytics, and it’s a worthy career in its own right.
The skills gap: what to learn before you apply
Most analysts have 70% of the skills already. The 30% gap breaks into four specific pieces, in roughly descending importance:
1. dbt (or equivalent)
The tool is almost always dbt. Learn the mental model: source tables → staging models (one per source, lightly cleaned) → intermediate models (business logic) → mart models (the thing analysts query). Learn the macros. Learn to write tests in schema.yml. The "dbt Fundamentals" course is free and is the hiring manager’s expected baseline.
2. Git and code review
If your analyst workflow is "SQL in a notebook, commit nothing," this is the biggest jump. Branches, PRs, commit hygiene, reviewing others’ code. A small personal dbt project on GitHub with real PRs against yourself is a credible portfolio signal.
3. Dimensional modeling
You know star schemas from querying them; now you have to design them. Kimball’s "The Data Warehouse Toolkit" is the classic; skim it for the vocabulary (conformed dimensions, SCD types, factless fact tables) so you can use those words in interviews.
4. Warehouse-specific quirks
Snowflake’s COPY INTO and warehouse-sizing semantics. BigQuery’s partitioning and clustering. Redshift’s sort/dist keys. You don’t need to be expert — you need to know the vocabulary of the platform the team you’re interviewing at uses.
Pay, title, trade-offs
The honest version of the numbers:
- Base salary uplift. AE roles typically pay 10–20% more than equivalent-level analyst roles, because of the SWE-adjacent skill set. The uplift compresses at senior levels — a senior AE and a senior analyst at the same company often land in the same band.
- Title. Entry-level AE roles are uncommon — the role usually sits at a 2+ year floor. If you’re a 1-year analyst, you’re looking at "analytics engineer" not "associate analytics engineer." Your AE title will read slightly senior to your analyst title.
- What you give up. Stakeholder exposure, ad-hoc exploration time, presentation opportunities. The role is more code, less slide.
- What you gain. Better tooling, clearer ownership, more engineering discipline, a portable resume line ("dbt + Snowflake" is a hiring magnet right now).
Analytics engineering isn’t a universally better job than analytics. It’s a specific career shape for people who want more code and less Slack — and the jump is shorter than most analysts think once you’ve shipped one real dbt project.