All categories/Data and analysis
Data and analysis
Definitions, extraction, anomalies, SQL. Have it restate the definition and most rework vanishes.
50 entries here · one process per category, checkpoints written per job
- Metric definition: nail the definition firstTwo people report different numbers for the same metric; usually the definition is missing, not the data.
- Natural language to SQL: schema firstYou have tables and a business question, and you do not want invented column names.
- Anomaly: hypotheses before investigationA number jumps or drops and the first instinct is usually the wrong place to look.
- Metric definition: agree the definition firstThree people gave three answers for monthly active users. We need one definition.
- Anomaly: the number halved yesterdayA core metric dropped 40% overnight. I need an investigation order, not a guess.
- Funnel: where the drop is worstThe signup flow has five steps and conversion is poor. Find which step is the problem.
- Retention: what a day-7 cliff meansDay-7 retention is 5%. Is it product, channel, or expectation?
- A/B readout: when the result is not significantTwo weeks in, variant A is 2% above B but not significant. Can we ship it?
- Segmentation: stop trusting the averageAggregate data looks flat while the business feels wrong. The average may be hiding it.
- Attribution: channels taking creditFour channels each claim the conversions, and the totals do not add up.
- YoY vs MoM: which one answers the questionThe report shows both year-on-year and month-on-month, and they contradict. Explain when each applies.
- Data cleaning: which rows can goAn export has duplicates, gaps and messy formats. I need rules, not silent deletion.
- Missing values: a gap beats a wrong numberA key field is 30% missing and the business wants it filled before analysis.
- Metric system: build it from scratchA new business needs a metric system: one view, no duplication.
- Report design: does this need actionThe daily report gets one reaction: too much, do not know where to look.
- Dashboard: a different first screen per roleOne dashboard serves ops, sales and the boss, and satisfies none.
- Competitor comparison: mismatched definitionsCompetitor data uses different definitions. Placing them side by side misleads.
- Price sensitivity: how much before they leaveEvaluating a 5% price rise with no experiment, only order history.
- Forecasting: no fake precisionA next-quarter sales forecast is needed. The boss wants one number; history is volatile.
- Sampling and surveys: is the sample representativeA survey returned 800 responses. Can the conclusions be generalised?
- Data story: three numbers into one linePresenting findings to a non-data audience. Scatter plots lose them.
- Explaining a decline: three causes, one testThe boss asks why the metric fell. I need an answer that is not a guess.
- Data quality monitoring: knowing it brokeThe business finds data problems before we do. We need to know first.
- Revenue decomposition: explainable piecesRevenue rose 12%. Which parts lifted it and which dragged?
- Weekly numbers: thirty secondsThe weekly meeting gets five minutes of table reading and nobody remembers any of it.
- Defending a finding: how to answerMy finding was rejected on the spot as 'wrong data'. I need a way to respond.
- One query: from question to resultThey want paying users per channel for the last 30 days. I need a query that runs.
- Slow query: read the plan firstA reporting query takes eight minutes. Target: under one.
- Joins: avoiding row multiplicationAfter joining three tables the row count grew and the result is untrustworthy.
- Window functions: ranks and running totalsRanks per channel per day plus running totals. I am unsure which construct to use.
- Distinct counts: how many people reallyUser counts differ because one person has many rows. Clarify the deduplication.
- Retention SQL: cohort by signup weekCompute next-day, day-7 and day-30 retention per signup week in one reusable query.
- Funnel SQL: five steps in one queryFive signup steps with strict ordering. Compute conversion in one query.
- Bucketing SQL: spend tiersBucket users into five spend tiers with counts and revenue share.
- Validation SQL: make two numbers reconcileThe report total does not match the sum of details. I need reconciliation queries.
- Schema design: think about the queries firstDesigning an order-events table: fields and indexes.
- Index advice: which columnA table got slow. Which index should be added?
- ETL: daily incremental logicA daily incremental job needs idempotency and backfill.
- Tracking plan: questions before eventsProduct needs tracking and engineering wants a plan, not a pile of useless events.
- Data dictionary: no guessing field meaningInheriting a database of hundreds of uncommented fields. Build a usable dictionary.
- Definition map: one word, two algorithmsOps and finance define GMV differently, and the reports fight every month.
- SQL debugging: right syntax, wrong resultThe query runs but returns nothing, or too little.
- Large table scans: read lessA query scans billions of rows. Reduce the read volume.
- Backfill script: several days missingA failed job left three days missing, and the table is overwritten in place.
- Sampling check: full scan too expensiveTo verify logic, but a full run costs two hours.
- Daily automation: query to deliveryThe same queries run manually each day and pasted into chat. Automate it.
- Query handover: fewer follow-up questionsAfter receiving the numbers they keep asking how they were computed. Attach a note.
- Time windows: calendar month or rollingCalendar month and rolling thirty days differ a lot, and the report got challenged.
- Report access: who sees whatA report with customer detail must serve several teams. Design the access.
- Data request intake: ask what it is forThe request is one line asking for a table, and nobody uses the result.
Try another category
- Writing and copy47
- Office and documents52
- Code and debugging50
- Learning and exams47
- Life and decisions53
- Agents and automation48
- Image and video45
- Job hunting30
- Workplace communication39
- Marketing and growth36
- E-commerce36
- Teaching and training36
- Research35
- Translation and localisation28
- Legal and contracts35
- Finance and operations35
- Product and design39
- Health35
- Customer support48
- Hiring and people ops48
- Social and content ops47
- Engineering and manufacturing48
- Security and operations48