Topic
Data & Decisions
Page 2 of 26 articles about Data & Decisions, newest first.
Browse the complete article archive →Filter Internal Traffic Without Removing Real Product Users
Test internal-traffic rules against staff and real-user cases, then distinguish GA4 Testing, permanent exclusion and a reversible report-only comparison.
Identity Merging Changed Your Analytics Counts: Check the User Mapping
Reconcile analytics events before and after identity mapping, with completed missed-link and incorrect-link cases that distinguish attribution from growth.
Late-Arriving Events Can Change Yesterday's Dashboard: Plan the Data Window
Explain changing daily totals with a worked late-event example, replacement snapshots, a completed recomputation policy and an explicit correction route.
NULL, Zero, and Missing Rows Mean Different Things in an Analysis
Separate NULL, zero and absent rows with a small SQL example. Compare averages, counts and coverage before replacing missing observations with zero.
Randomize Users, Accounts, or Teams? Choose the Unit Before the Experiment
Choose an experiment unit from shared product behavior, with a completed collaboration design, checked assignment maps, and explicit analysis limits.
Retention Rate Depends on the Cohort and Return Event You Choose
Define retention with a complete cohort, return event, window, and cutoff, then reconcile synthetic user records across exact-day and later-return measures.
Run a Data Backfill Without Double-Counting Previously Processed Records
Reconcile a fixed historical snapshot, reject conflicting duplicates, and verify partition failure recovery with an executed local backfill and rerun record.
Sample Ratio Mismatch in A/B Testing: Check Observed Counts Against the Planned Split
Check an A/B test against its planned allocation with a worked 70/30 SRM diagnosis, verified count calculations and a completed data-qualification record.
Session Counts and User Counts Tell Different Stories: Choose the Right Unit
Separate sessions, device keys, accounts and saved tasks. Audit a complete event mapping and correct a reporting memo without inventing new-user growth.
Time Zones and Daylight Saving Can Break a Daily Metric
Check daily metrics against named business time zones, 23- and 25-hour days, repeated local times, and a worked event reconciliation before changing reports.
Two Dashboards Disagree: Reconcile the Metric Definition Before Fixing the Chart
Reconcile conflicting dashboard totals with a worked definition sheet, record-level trace, timezone boundaries, filters, aggregation, and snapshot cutoffs.
Use Data Quality Checks That Match the Failure You Need to Catch
Match data quality rules to specific failures. Review a filled check matrix, thirteen executed synthetic cases and the blind spots a green result can hide.











