Reports can be perfectly formatted and wrong. A data-quality dashboard works like a traffic light: it checks whether conversions, costs, links and tags are within normal and warns before decisions are based on suspect numbers. It is one of the best-return tools in measurement.
The essentials in a definition
A data-quality dashboard is a traffic light that warns before reports lie: it checks whether events arrive, values are plausible, conversions match sales and links between systems work. It can be built in Looker Studio with data from events and sales sources, following NIST’s monitoring logic.
The problem with pretty numbers
Most data errors give no visible error: the dashboard keeps showing plausible values. A broken tag, a duplicated conversion or a stalled integration change the results without anyone noticing. Decisions are based on what seems reliable.
Data quality is not seen: it is measured.
What the dashboard checks
Whether conversions match, within a range, sales or leads in the CRM. Whether there are abnormal drops or spikes in events. Whether platform costs match invoices. Whether tags fire on the expected pages. Whether campaign identifiers reach the leads. Whether consent is being recorded. Whether there are duplicates.
Each check has a threshold and a colour: green, amber, red.
A data-quality traffic light
- Choose 5 to 10 essential checks
- Set thresholds from history
- Show status as green, amber and red
- Warn and record each problem
How to build it simply
Choose five to ten essential checks. Calculate them in a sheet or dashboard tool, with exported or connected data. Define reasonable thresholds from history. Show the state of each clearly, with the date of the last update. Send a warning when something turns red.
Start small and add with experience.
Before trusting the numbers, it helps to trust those who produce them.
Use and maintain
Consult the dashboard before meetings and important decisions. Record each problem found and its cause. Review thresholds periodically. Assign an owner. Share the status with the team and agencies, to create a culture of rigour.
When the light is green, you decide more calmly.
Decision table
| Check | Green signal | Red signal |
|---|---|---|
| Events arriving | Volume within normal | Sudden drop or zero |
| Conversions vs sales | Stable ratio | Ratio outside range |
| Values and currency | Plausible and consistent | Zero or extreme values |
| Source freshness | Data up to date | Delay longer than expected |
Common mistakes to avoid
- A dashboard only of results. Without verification, errors pass for facts.
- Fixed thresholds. Adjust by day and season.
- No owner. Someone must act when the light turns red.
How to measure the result
Record incidents detected, time to detection and fix, and false alarms, and improve thresholds.
Limits and exceptions
No dashboard detects all errors, and false alarms can cause fatigue. Some failures are subtle (slightly wrong values, partial duplication). Combine the dashboard with periodic audits and comparison with independent sources, and keep documentation of the checks.
Illustrative example
Illustrative example: a team creates a dashboard with five checks and email alerts. One morning, the purchase-event light turns red: a checkout change broke the event. The team fixes it in two hours, before optimisation learns from missing data. Without the dashboard, the problem would only be found in the monthly report.
In practice: checkpoints
Define a few quality checks, such as freshness of data, percentage of missing values, differences between platforms and the database, and duplicated events. Give each check a green, amber or red threshold, show the status at the top of every report and assign someone to act on reds. Review thresholds as the business changes, and keep a record of incidents and fixes. The main benefit is that decisions stop depending on numbers nobody trusts, and that problems are caught while they are still cheap to fix. Share the status with everyone who reads the reports, so that a red light changes how the data is used.
Frequently asked questions
Which checks matter most?
Conversions versus CRM, costs versus invoices and firing of the main tags.
Do I need an expensive tool?
No. A sheet or simple dashboard is enough to start.
Who maintains the dashboard?
Someone with clear responsibility, reviewed periodically.
Which checks should I include?
Events arriving, ratio between conversions and sales, plausible values and source freshness.
Who should monitor?
Whoever manages measurement, with a deputy, and automatic alerts so as not to depend on memory.
A data-quality dashboard with traffic lights (events, conversions vs sales, values and freshness) detects measurement errors before they distort decisions.
Sources and recommended reading
- Looker Studio for developers
- Analytics Help: GA4 recommended events
- NIST: AI Risk Management Framework
Last editorial review: October 2026. Informational content; always confirm current conditions in the official sources before deciding.