Hire a data engineer when the main problem is producing dependable, usable data. Hire a data analyst when the main problem is turning available data into sound answers for business decisions.
The roles overlap, especially in small teams. Define the work rather than assuming either title covers the entire data lifecycle.
Locate the blockage
| Problem | Likely emphasis |
|---|---|
| Sources arrive inconsistently | Data engineering |
| Transformations break or cannot be traced | Data engineering |
| Teams disagree on a business definition | Shared work with a business owner |
| Reliable data exists but decisions lack analysis | Data analysis |
| Reports are technically correct but misunderstood | Analysis and stakeholder communication |
| Nobody owns the dataset after delivery | Explicit operational ownership |
A dashboard request may hide a data production problem. A pipeline request may hide an unresolved business question.
Define the consumer
Ask who will use the data, for which decision, and how fresh or complete it must be. These requirements help the team decide what reliability means.
Do not build a broad data platform before understanding its consumers. Equally, do not ask an analyst to produce confident conclusions from a source whose meaning and quality are unknown.
Make the handoff explicit
For each important dataset, name the source owner, transformation owner, definition owner, and consumer. Some may be the same person initially, but the responsibilities should remain distinct.
Define how changes are communicated and how incorrect or late data is handled. A dataset delivered once is not necessarily an operated data product.
Assess the relevant reasoning
For engineering, use a fictional pipeline failure and ask how the candidate would trace, repair, and prevent it.
For analysis, provide a fictional decision question and a limited dataset description. Ask what can be concluded, what cannot, and what further evidence is needed.
In both cases, look for clarity about definitions and uncertainty.
Scope a combined role carefully
A combined hire needs priorities. State whether the early emphasis is foundations or decision support, and what work will wait.
Use the data quality interview and role overlap guide. Discuss a data hiring mandate with Refery.