Who & why
Answers business questions with data — turning raw numbers into decisions: which feature worked, where users leak, does a channel pay off. Not ML systems, not product code — gives teams a clear picture of reality instead of guesses. Without one: decisions by gut, contradictory metrics, misread A/B tests.
A day in the life
Morning: check dashboards for anomalies, take a product question. Day: write SQL, clean data, segment to find the cause, compute A/B significance, build dashboards. Evening: phrase the conclusion in plain language, fix metric definitions.
Key skills
Hard: advanced SQL, dashboards (Metabase/Tableau), product analytics (AARRR/HEART, funnels, retention), A/B statistics, data cleaning, basic Python (pandas). Soft: critical thinking (correlation ≠ causation), data storytelling, decision-oriented conclusions, skepticism of pretty numbers.
Artifacts
Dashboard, SQL queries, A/B report, metric definitions. Builds on the Product metrics & Unit economics methodologies.
How AI / vibe-coding boosts the role
SQL generation & explanation; anomaly detection; A/B interpretation; metric definition; funnel analysis — with ready prompts.
Growth: Junior → Middle → Senior → Lead
Junior: SQL & dashboards to a task. Middle: owns an area's analytics. Senior: complex data research, influences product. Lead/Head: analytics function & data culture.
Common mistakes
Correlation = causation; metrics without shared definitions; eyeballing A/B; a chart with no conclusion; dirty data → pretty report.
What to learn
Product analytics (AARRR/HEART), cohorts & retention, A/B statistics, correlation vs causation, funnels. Read: Storytelling with Data; Trustworthy Online Controlled Experiments.
Salary (RU)
Junior ~80–140k₽/mo, Middle ~140–230k, Senior ~230–350k. Varies by grade/industry/city/year — check current data.
Laskoff agent mapping
No direct mapsTo; the data/metrics loop is closest to the analyst agent. Home skill: product-tracking. Unit economics can be computed in the service's built-in calculator.
🤖 Persona prompt
You are an experienced Data Analyst. Help me answer business questions with data, not gut feeling. Always clarify which decision depends on the answer. Write clean, documented SQL and explain the logic; warn about data pitfalls (dupes, NULLs, timezones). Evaluate A/B statistically (significance, power), not by eye. Clearly separate correlation from causation. End every analysis with a plain-language conclusion and an action recommendation. Suggest shared metric definitions so everyone counts the same.