Corrupted data on my dashboard: what to do
Understand what corrupted data means in Metrikia, how it appears, and what steps to follow to request a cleanup from support.
What is corrupted data
"Corrupted data" refers to rows imported into Metrikia whose import date is earlier than a known cutoff date. This typically happens after:
- A bulk CRM import that injected historical data into recent periods
- A misconfigured pixel or connector before it was fixed
- An accidental re-import of leads or deals already present
These rows can produce anomalous metrics: profit 4x spend, show-up rate at 100%, deals won "yesterday" when nothing was actually sold.
How Metrikia detects the issue
Metrikia automatically compares the actual date of each record (as reported by your CRM or Stripe) with the date it was imported into Metrikia (imported_at). When the gap exceeds a threshold, the record is flagged as potentially historical.
The dashboard displays a warning banner when the proportion of historical data is high.
What to do if your metrics look wrong
- First check the selected period: a bulk import on "yesterday" inflates all of that day's metrics
- If metrics remain anomalous, contact Metrikia support via /app/support/diana or by email
- Provide: the affected period, the impacted metrics (profit, show-up, leads...), and the approximate date of the problematic import
A Metrikia administrator will analyze your data and perform the cleanup if needed.
After the cleanup
Once the data is purged, re-trigger synchronisation from:
- /app/settings > Integrations > your CRM connector > Sync now
Metrics recalculate automatically after the sync.
Common issues
- "My profit shows 400% ROAS": contact support, this is the classic sign of a historical import polluting metrics.
- "The banner disappears then comes back": check that your CRM connector is not re-sending the same historical data on every sync. Support can block the re-import.
- "I want to wipe all my data and start over": this is not a user-level action. Contact support with your explicit request.
To go further, check out our blog, the documentation or contact support.