Data Scientist / Analyst Job Search Playbook
A 2026 guide for data roles — the analyst-to-DS fork, SQL test prep, the "how would you measure X" interview, and portfolio essentials.
Analyst vs. Data Scientist in 2026
Analyst roles are about business metrics + dashboards + stakeholder communication. DS roles are about experimentation + modeling + statistical rigor. The boundary is blurry, and companies use titles inconsistently. What matters: if you can't write a stats-sound A/B test analysis, you're Analyst. If you can, but can't ship a production ML model, you're DS (non-ML). If you can do both, you're senior DS — charge accordingly.
The SQL test
Most data interviews open with 1-2 SQL problems. The common trap is window functions — ROW_NUMBER, LAG/LEAD, running aggregates over partitions. Practice these cold. A clean "count the N-th event per user" solution signals seniority better than any resume bullet.
The "how would you measure X" interview
Classic opener: "How would you measure the success of [feature]?" The answer is not a number — it's a framework. Primary metric (what we're trying to move), guardrail metrics (what can't get worse), segmentation (which users), experimental design (A/B vs. DID vs. observational). Walk through the framework out loud; the interviewer will push on tradeoffs.
Portfolio — what actually gets opened
GitHub notebooks get ignored. What gets read: a public analysis write-up (1500 words max) on a real dataset that answered a question someone would actually ask. "I analyzed Lyft's public pricing data and found X" lands better than 10 Kaggle notebooks. Blog it.
ML engineer vs. DS compensation
If your role involves shipping ML models to production, push hard for the ML Engineer title — comp is ~$40-60k higher at senior levels. "Data Scientist" roles are increasingly analyst-comp. "ML Engineer" is engineer-comp. Titles matter more than scope here, even for identical work.
Put this playbook on autopilot.
Incognito scans, matches, and applies so you can focus on the interview prep this guide actually teaches.
Try it free →