How does a data analyst's work differ from that of a data scientist and a data engineer?
The three roles differ mainly by where they sit in the data flow and which question they answer. A crisp way to frame it: the engineer makes data available, the analyst explains what happened and why, and the scientist estimates what will happen next.
Data engineer — builds and operates ingestion pipelines, warehouse schemas and orchestration; optimises for reliability, freshness and cost, and works mostly in SQL, Python and pipeline or cloud tooling.
Data analyst — answers descriptive and diagnostic questions (what happened, where, and why), builds dashboards and semantic models, and converts findings into business recommendations.
Data scientist — answers predictive and prescriptive questions, designs experiments, and builds statistical or machine-learning models, with more emphasis on modelling technique and validation.
Shared ground — all three write SQL and Python and care about data quality; the split is about depth and ownership, not exclusive skills, and job titles vary widely between companies.
The middle role — analytics engineer has emerged between engineer and analyst, owning transformation and modelling so analysts get clean, tested, well-documented tables.
Key terms: descriptive analytics, diagnostic analytics, predictive modelling, data pipeline, orchestration, semantic model, analytics engineer, experimentation
Give the one-line split first (available / what happened / what will happen), then immediately say the boundaries blur by company size — that shows you have worked in a real team rather than read a diagram. Interviewers often use this to probe ambition, so if you are asked which you want to become, tie the answer to the role you are applying for and to concrete work you enjoy.
Never disparage the other roles or describe analyst work as a stepping stone to data science; that reads as low commitment. A good closer is naming the analytics-engineer role, which shows you are current on how teams are actually structured.