A realistic workday map
FrameDefine the decisionTranslate the request, identify the metric, name assumptions, and find the real question.
InspectFind the data shapeTables, joins, missingness, definitions, historical changes, and whether the source can be trusted.
BuildQuery or dashboardSQL, spreadsheet, BI view, validation checks, and chart choices.
ExplainTurn it into a decisionMemo, deck, chart notes, caveats, recommendation, and the argument for what to do next.
Follow-upKeep it honestAnswer questions, revise assumptions, monitor changes, and document the next version.
What to watch when you shadow
Watch how long it takes before the person touches the final tool. The strongest data workers do not rush straight to a chart or model. They ask what the decision is, who will use the answer, what could be wrong with the data, and what evidence would change the recommendation.
Request qualityIs the ask a real question, or just a request for a number?
Data qualityHow much work happens before analysis because the source data is messy?
Review cultureDoes anyone check definitions, assumptions, or model validity?
Decision linkDoes the work change what the team does next?
If the day looks quiet, do not mistake that for easy. The hard work is often mental: resisting a false answer, naming uncertainty, and keeping the analysis useful when the data refuses to be clean.
How the day changes by company maturity
At an early startup, the day may be half analytics and half data janitor: events missing, tables changing, founders asking urgent questions, and nobody fully owning the data model. At a mature company, the day may involve more governance, review, permissions, metric councils, experimentation standards, and coordination with data engineering. Neither version is automatically better. They reward different temperaments.
StartupMessy
More ambiguity, more impact
You may get closer to decisions quickly, but you also inherit weak instrumentation and rough process.
EnterpriseLayered
More systems, more politics
Definitions, permissions, governance, and ownership can protect quality or slow the work down.
Product-ledFast
More experiment rhythm
Questions revolve around user behavior, funnels, feature adoption, retention, and product tradeoffs.
RegulatedCareful
More audit trail
Healthcare, finance, insurance, and government data work puts more weight on privacy, risk, and defensibility.
Sources and methodology
O*NET Database 30.3Occupation descriptions, alternate titles, work context, work activities, and education signals.
BLS OEWS May 2025National wage estimates, percentile pay, mean pay, and employment estimates by SOC group.
BLS Employment Projections2024 to 2034 projected employment, growth, annual openings, entry education, experience, and training.
BLS OOH profileOfficial Occupational Outlook Handbook context for the matched career family.
Career Dish adds fit scores, workload metrics, AI exposure estimates, and interview-style guide scenes on top of public datasets. Those interpretive layers are meant to make the data scannable, not to replace official licensing or school-specific research.