Products Solutions Templates Pricing Resources

Setting up alerts for data quality issues: preventing bad data

Home / Everything About / Everything About Analytics / Setting up alerts for data quality issues: preventing bad data

Your conversion tracking breaks at 2 PM. You don't notice until 4 PM when you check your dashboard. Two hours of conversions are missing. An alert system would have notified you immediately.

Alerts are automated notifications that trigger when something goes wrong. They turn data quality from reactive (you notice a problem) to proactive (the system tells you).

Why alerts matter

Speed of detection

The faster you know about a problem, the sooner you can fix it. An alert tells you immediately instead of waiting hours for your next check.

Reduces impact

The shorter a tracking outage lasts, the less data you lose. An alert that wakes you up at 2 AM costs sleep but saves the day's data.

Catches automation errors

If you have automated processes that change configuration or data, alerts catch when those go wrong. Without alerts, automated errors compound.

What to alert on

Traffic volume anomalies

Alert if today's traffic is 50% below the weekly average. Something broke or changed significantly. This is the first sign of tracking problems.

Conversion count anomalies

Alert if conversions drop suddenly without corresponding traffic drop. Conversion tracking might be broken.

Conversion rate anomalies

Alert if conversion rate changes dramatically. This might indicate a real problem or a tracking issue.

Analytics vs. payment processor variance

Alert if variance between analytics and payment processor exceeds your normal range. If they normally match within 5% but suddenly 20% off, something's wrong.

Bot traffic spikes

Alert if bot percentage suddenly jumps. You're under attack or bot filtering broke.

Configuration changes

Alert when settings change. When a filter is added or removed. When a goal is created or disabled. Someone changed something.

Missing data

Alert if data stops being collected for a specific source or page. Event tracking might be broken for that area.

Data freshness

Alert if data hasn't been updated in 24+ hours. Reporting might be broken.

Setting up alerts

Use your analytics tool's built-in alerts

Google Analytics has email alerts. Create an alert for traffic drop. Create one for conversion drop. These require no setup beyond the alert creation.

Use a BI tool's alerting

Tools like Tableau, Looker, and Google Data Studio have alert functionality. Define metrics. Set thresholds. Trigger alerts via email.

Use third-party monitoring tools

Services like Sentry, Datadog, or custom monitoring can track analytics health. These integrate with your analytics API and alert on issues.

Build custom monitoring

Pull metrics from your analytics API daily. Compare to thresholds. Send email or Slack notification if thresholds are exceeded. Requires development but offers full customization.

Alert best practices

Set thresholds realistically

Too aggressive and you get alert fatigue (ignoring real alerts). Too lenient and you miss real problems. Start conservative. Adjust based on false positives.

Alert to the right person

Traffic alerts go to the analytics owner. Conversion alerts go to the marketing team. Configuration alerts go to the tech team. Right person gets the message.

Include actionable information

"Traffic dropped 50%" isn't actionable. "Traffic dropped from 1000 to 500 yesterday afternoon. Check if tracking code is firing." is actionable. Include context.

Test alerts

Create a test alert. Trigger it manually. Make sure you receive the notification. Verify it includes the right information. Don't rely on an alert until you've tested it.

Regular review of alert rules

As your business grows, alert thresholds need adjustment. Monthly, review alert performance. Are you getting too many false positives? Adjust. Are important issues not alerting? Add them.