Health alerts that tell you when the signal breaks, and stop when it is back
Convrail checks every store every 15 minutes with four monitors: event volume drop, Conversions API batches that failed permanently, a pixel that stopped sending while the server still does, and failed feed deliveries. Each problem opens exactly one alert, in the app and by email, and resolves itself when the condition clears.
The four monitors and their thresholds
Thresholds are fixed and documented so an alert always means the same thing. They are computed from your own store’s history, not from a global average.
| Monitor | Condition | Severity |
|---|---|---|
| Volume drop | Events in the last 24 hours below 30% of the daily baseline (average of the previous 7 days), when the baseline is at least 50 events | critical |
| Conversions API errors | At least one batch failed permanently in the last hour after all retries | critical |
| Pixel broken | Zero browser events in the last 24 hours while server events keep arriving | warning |
| Feed delivery failed | At least one feed run failed to deliver in the last 24 hours | critical |
One alert per problem, never a flood
A monitor that runs every 15 minutes would send 96 emails a day if it were naive. Convrail keeps at most one open alert per store and per type. The alert is created once, the email is sent once at opening, and the alert is marked resolved automatically when the monitor stops firing. If the problem comes back later, a new alert opens.
Detect
The monitor evaluates the four conditions from your event, batch and feed run counters.
Open once
If a condition is true and no alert of that type is open, one alert is created with the exact message and counters.
Notify
If notifications are enabled, one email goes to the address you set. The alert is also listed on the Health screen.
Resolve
When the condition clears, the open alert is resolved with a timestamp. No email is needed to tell you it is over: the Health screen shows it.
Weekly report and exports
If you opt in, a weekly email summarizes the previous 7 days every Monday at 08:00 UTC: attributed revenue split organic and ads, spend and ROAS when the Advertiser API is connected, events processed, feed runs and deliveries. The attributed orders of any period can be exported as CSV from the Health screen.
- Notification settings live in the app: alerts on or off, weekly report on or off, one email address.
- Emails are plain text and contain the alert ID so you can find it in the app.
- Nothing is sent when you have not set an address; alerts still appear in the app.
Frequently asked questions
How fast will I know that my pixel stopped working?
The pixel monitor compares browser and server events over a rolling 24-hour window, evaluated every 15 minutes. A pixel that goes silent while orders keep arriving server-side is flagged as soon as the last browser event is older than 24 hours.
Will I get an email every 15 minutes while the problem lasts?
No. One alert per store and per type is open at a time. You receive one email when it opens; it resolves itself when the condition clears.
What does "volume drop" compare against?
Your own daily baseline: the average number of events per day over the previous 7 days. The alert fires when the last 24 hours are under 30% of that baseline, and only once the baseline reaches 50 events a day so small stores are not flooded with noise.
Can alerts go to several people?
Notification settings hold one address per store today. Use a shared mailbox or a distribution list if several people need them.
Related guides
Measure the ChatGPT channel this week
Free during early access while we onboard the first stores. Leave your email and we send the install link when your spot opens.