SI Signal & Funnel
Measurement Foundations

Marketing Data Quality Checklist Before Analysis

Signal & Funnel · Updated 2026-07-19 · 4 min read
Marketing Data Quality Checklist Before Analysis
SummaryCheck marketing data quality against a defined question and metric. Test normal, invalid, duplicate, canceled, and consent-denied collection paths; inspect freshness, completeness, allowed values, time zones, currencies, and taxonomy; reconcile systems with documented differences; and count rows around every join. Mark missing, delayed, and zero separately, preserve version changes, protect identity and access, and state which conclusions remain usable before anyone acts on the report.
Contents

Freeze the question before checking the data

A marketing data-quality check begins with the metric definition, decision, scope, and expected source. You cannot validate “leads” until you know whether the report means form submissions, accepted records, qualified prospects, or accounts. Quality means fitness for the stated use.

Use the conversion guide when the event or eligible denominator remains unclear.

Confirm that required tags, events, imports, and server or offline handoffs fire only in the intended conditions. Test valid, invalid, duplicate, canceled, and consent-denied paths where applicable. Verify environments, domains, app versions, and time zones.

Collection must comply with current privacy, consent, security, contract, platform, and sector duties. Never disable or bypass a person's choice to make a chart look complete. Use qualified local legal, privacy, and security guidance for consequential systems.

Inspect completeness and freshness

Compare expected records with received records by source and period. Mark missing, delayed, and not applicable separately from zero. Check the last successful load, normal processing delay, and whether a partial day or month is being compared with a complete one.

Investigate sudden perfect round numbers and flat lines. They may reflect real stability or a failed data connection.

Test validity and allowed values

Check types, ranges, required fields, currency, timestamp, campaign taxonomy, URL, market, and event sequence. Values outside the documented dictionary need review rather than automatic reassignment.

The UTM guide helps validate source, medium, and campaign labels without placing personal information in URLs.

Reconcile across systems

Compare source totals at defined handoff points. Expect legitimate differences from time zones, attribution windows, identity rules, refunds, late validation, consent, and platform processing. Document them.

Do not force two systems to match by deleting inconvenient records or silently changing the window. Reconciliation explains differences; it does not guarantee identical numbers.

Check joins and identity

Count rows before and after joins, test key uniqueness, inspect unmatched records, and look for many-to-many duplication. Confirm whether the unit is a person, browser, account, session, order, or event.

Identity matching can be probabilistic and privacy-sensitive. Minimize data, restrict access, and do not treat device identifiers as certain people. Never connect datasets in ways that violate law, consent, contracts, or platform terms.

Review consistency and version changes

Compare definitions, filters, taxonomy, and event logic across periods. Annotate releases, consent changes, new campaigns, migrations, and backfills. Preserve the original and transformed fields where governance permits so changes remain traceable.

Use the dashboard guide to display data status and breaks rather than smoothing them away.

Record the decision and owner

Classify issues by impact: blocks reporting, limits a segment, affects only presentation, or requires monitoring. Name the owner, correction, validation test, and backfill decision. Do not backfill unless the method is documented and the result remains distinguishable from directly observed data.

Finish with what is usable, what is not, and which conclusions must wait. Do not let a consequential decision proceed on data that failed the relevant checks.

Keep a release record

For every approved dataset, record the test date, scope, owner, exceptions, affected metrics, and next review. A later analyst should be able to tell which checks ran and which did not. Reopen the record when tracking, consent flows, identity logic, source fields, or joins change.

Official rule sources

Data-protection and direct-marketing duties depend on jurisdiction, data, purpose, and message. Check the current official source relevant to the people and activity: the European Commission data-protection portal for EU scope, the UK Information Commissioner's Office direct-marketing guidance updated 28 April 2026, the California Privacy Protection Agency laws and regulations for California scope, and the U.S. Federal Trade Commission CAN-SPAM guide for U.S. commercial email. These official pages do not determine whether a rule applies to a specific business. Also check current platform documentation and contracts, and use qualified local privacy or legal counsel for consequential decisions.

General marketing education, not legal, privacy, tax, financial, security, or individualized business advice. An independent publication. Not affiliated with any prior owner of this domain.

FAQ

Why do marketing platforms show different totals?

Differences can come from attribution windows, time zones, identity rules, consent, blocked collection, refunds, processing delays, filters, and event definitions. Align scope and definitions, compare at a documented handoff, and quantify the remaining gap. Do not force equality by deleting records or changing rules silently. Each system may be answering a different question even when the labels look similar.

How do I distinguish zero from missing data?

Zero means the system observed no qualifying value within the defined scope. Missing means the value was not received, not available, or not applicable. Store and display these states separately, monitor freshness, and label partial periods. Replacing missing data with zero can create false drops, rates, and decisions. Define how each state behaves in calculations before reporting.

Can I backfill missing marketing data?

Only when a documented, defensible method and governance process permit it. Preserve the distinction between directly observed, reconstructed, and estimated values; state assumptions and affected periods; and validate the result. Do not manufacture events or overwrite source history to make a trend continuous. Personal-data backfills must also satisfy current privacy, consent, security, contractual, and legal requirements.

Which official privacy and marketing sources should I check?

Use the official source that matches the people, jurisdiction, data, and activity: the European Commission data-protection portal for EU scope, the UK Information Commissioner's Office direct-marketing guidance for UK scope, California Privacy Protection Agency laws and regulations for California scope, and the U.S. Federal Trade Commission CAN-SPAM guide for U.S. commercial email. Then check current platform documentation and contracts. Qualified local counsel should review consequential decisions.