Study first
Review the ideas behind the questions
Review the checks that make campaign data safe to combine. Focus on missing records, duplicates, consistent definitions, freshness, valid formats, accuracy, metadata, and clear caveats.
Check The Data Shape
Before numbers are combined, the team should know whether expected records exist, duplicate rows are controlled, and values use the expected format.
- Completeness asks whether expected records and important values are present.
- Uniqueness asks whether each entity appears once instead of being duplicated.
- Validity asks whether data is in the expected range and format.
In Practice
Missing Data Changes The Claim
If a required channel file or conversion value is missing, the report should say what is absent before comparing totals.
Duplicates Inflate Counts
A report that counts the same lead or order twice can overstate outcomes even when the total row count looks complete.
Common mistakes
Treating a full export as trustworthy without checking duplicates or required fields.
Check completeness and uniqueness separately because a data set can have rows and still duplicate entities or miss important values.
Q&A
Can a complete file still have a data-quality problem?
Yes. It can contain expected rows but still include duplicate records, invalid formats, or inaccurate values.
What should be checked before counting outcomes?
Check whether the outcome records are present, unique, valid, and tied to the same definition.
Make Definitions Comparable
Cross-channel reporting is weaker when each source uses a different meaning for exposure, engagement, outcome, or time period.
- Consistency asks whether values in one data set contradict values representing the same entity in another data set.
- Shared standards improve comparability when performance definitions differ across media.
- Metric definitions should be clear enough for users who do not already know the data-quality terms.
In Practice
Define The Same Outcome
Before adding channel results, confirm that each source counts the same outcome in the same period.
Explain Definition Changes
If a lead, order, or engagement definition changed, the trend should name that change before readers compare periods.
Common mistakes
Adding channel totals before checking whether each channel counts the same outcome.
Define the outcome and time period first, then combine numbers that are comparable.
Q&A
Why does a metric glossary matter in a dashboard?
It helps readers understand what each metric means and whether numbers from different sources can be compared.
Communicate Quality Limits
A data issue does not always block a report, but the limit should be visible enough for readers to use the numbers properly.
- Timeliness depends on whether the data reflects the period it represents and is available soon enough for the intended use.
- Accuracy asks whether data matches reality, and users should be told about bias that may affect quality.
- Transparent and quality-assured metrics are easier to audit and trust.
In Practice
Show Freshness
If one source is delayed, label the data period and explain whether the result is provisional.
State Caveats Plainly
A caveat should explain how the issue affects use of the report, not hide inside technical notes.
Common mistakes
Treating stale or biased data as final because the dashboard has already refreshed.
Check whether the data is timely and accurate enough for the decision, then disclose any caveat that affects use.
Q&A
What should a report say when data is provisional?
It should label the data period, explain the delay or quality trade-off, and say how that affects use of the result.
Why is auditability part of metric quality?
A metric is stronger when its method and checks are transparent enough for others to review.