Data Modeling Before Dashboards: Build Metrics People Can Trust
Use business definitions, entities, events, and trusted marts before investing in dashboard polish.
Data Foundations
How to shape raw data into documented tables, shared definitions, and durable business concepts.
Use business definitions, entities, events, and trusted marts before investing in dashboard polish.
A practical introduction to using the data warehouse as the center of your analytics system.
A practical guide to deciding who owns data work, what a runbook should contain, and how to keep data systems reliable after the first build.
How to move old reports into a modern data stack without losing metric trust, business context, or operational continuity.
How to pass business questions, metrics, models, and ownership from one team or system to another without losing trust.
Why a successful pipeline run does not always mean the data is current, and how to model freshness so dashboards stay trustworthy.
The practical mistake that causes historical data repairs to create new trust problems instead of fixing old ones.
Why most customer models fail by mixing people, accounts, subscriptions, and events into one unstable definition.
A practical checklist for defining metrics clearly enough that dashboards, data models, and business conversations stay aligned.
A practical checklist for spotting, triaging, and controlling changes in source systems before they damage models, pipelines, and dashboards.
A practical checklist for turning messy operational data into data that analytics, automation, and AI systems can safely use.
A practical way for founders to decide when the warehouse should become the center of reporting, modeling, and business measurement.
A practical way to move old reports into a trusted data model without breaking the numbers the business still runs on.
A practical way for founders to make data systems accountable, recoverable, and less dependent on memory.
A practical way to move analytics out of the founder's head and into a trusted operating system.
A practical migration plan for making stale data visible, measurable, and fixable before users lose trust in the system.
A practical beginner playbook for moving, rebuilding, or repairing historical data without breaking trust in the new model.
A practical way to redesign customer entities, identifiers, and history before migrating dashboards, pipelines, or CRM reporting.
A practical note on making business metrics clear, testable, and stable enough for dashboards and decisions.
How small changes in operational systems quietly break models, pipelines, and dashboard trust.
A practical guide to separating possible row identifiers from the one you choose as the primary key.
Use dependency violations to find attributes stored at the wrong grain before they damage metrics and dashboards.
A practical guide to deciding when several columns together identify one business event.
How to explain identifier stability when a source system identifier can change, be reused, or stop meaning what it used to mean.
How to find orphaned facts while preserving legitimate late-arriving dimensions.
A practical guide to why UNKNOWN changes SQL filters, joins, constraints, and predicate reasoning.
A practical guide to showing why duplicate rows survive normal SQL queries unless you explicitly remove or prevent them.
How to reshape repeated attributes into child rows while keeping the original parent record clear and queryable.
A practical guide to separating partial and transitive dependencies without turning normalization into theory theater.
A practical guide to proving whether decomposed tables join back to the original records without missing or invented rows.
A practical guide to separating a business thing from the many rows that describe, observe, or mention it.
How to model records whose identity only makes sense inside a parent record.
How to model relationship attributes without flattening two entity tables into a misleading shape.
A practical way to find entities that satisfy every requirement in a reference set, not just some of them.
A practical guide to deciding when two records may share the same identifier without describing the same thing at the same time.
A practical guide to sizing decimal columns and calculation outputs so stored values, sums, products, and ratios do not overflow or silently lose useful detail.
How to store original units, normalized values, and conversion provenance without losing numeric meaning.
How to tell whether a number should use the transaction date rate or the reporting date rate before it reaches a dashboard.
A practical guide to preventing multiplication-versus-division mistakes in currency normalization.
A worked comparison for avoiding absolute-versus-relative change errors in metrics, models, and dashboards.
How to explain estimated unique counts honestly, set acceptable error, and avoid presenting probabilistic metrics as exact facts.
How to combine group statistics correctly without averaging incompatible variances.
How to spot metrics where arithmetic averages hide the real cumulative change.
How to keep long numeric identifiers from being rounded as they move through files, APIs, warehouses, spreadsheets, and BI tools.
How to model unavailable observations separately from real absence of activity so totals, averages, rates, and alerts stay honest.
Use start-inclusive, end-exclusive boundaries to classify timestamped events once and only once.
How to model fiscal boundaries without assuming months, quarters, or years line up with the calendar.
How to explain why weekly reports around New Year can appear to move days into the “wrong” year.
A practical guide to telling clock disagreement apart from real event reordering in multi-system logs.
A practical guide to separating business validity from system knowledge in data models and reports.
Represent corrected history without losing what your data system believed at the time.
A practical guide to separating hours elapsed from calendar days advanced in data models, SLAs, and time-based metrics.
A practical guide to designing event-level fact tables that reconcile cleanly and do not mix transactions with rollups.
A practical guide to designing snapshot facts for balances observed at regular intervals, including days with no business activity.
A practical guide to modeling processes such as orders, claims, tickets, and applications as one row that updates as milestones are completed.
How to model who or what was expected to participate separately from the events that actually happened.
A practical way to state which dimensions permit summation before a metric reaches a dashboard.
How to keep invoice, order, ticket, or receipt identifiers in a fact table without building empty dimensions.
A practical dimensional modeling pattern for assigning one amount to several recipients without inflating totals.