Ranked by what it costs
Every finding is sized in lost completions: flows that would have completed without the problem, measured against comparable flows on the same step, device and browser. The biggest loss sits at the top of the list.
An error list tells you that something fails. Findings tell you what is going wrong, one root cause at a time: who has to act, how many logins it costs and how to fix it.
One root cause per finding, sized in lost completions
Fix options with a prompt for your coding agent
Evidence, affected flows and a status timeline
Impact. Owner. Fix. Status.
Each finding is a short, complete statement about one problem, so the people who have to fix it do not have to rebuild the analysis first.
Every finding is sized in lost completions: flows that would have completed without the problem, measured against comparable flows on the same step, device and browser. The biggest loss sits at the top of the list.
Each finding names its type (defect, incident, friction, platform limitation or expected behaviour) and its owner: your frontend, your backend or IdP, the edge or CDN, a third party or the end user.
Findings come with fix options, the recommended one marked. Each option carries a ready prompt for your coding agent, so the team that owns the problem can start right away.
See which explanations were checked and whether each one is confirmed, ruled out or still open. The evidence tables show the data behind the answer and the request that produced it.
When users simply cancel a prompt, the finding is marked as expected behaviour and shows how often it happens instead of a loss, so it never competes with real defects for attention.
Set a finding to in progress, fix shipped or accepted. Corbado then re-measures the impact and marks the fix verified. The timeline keeps every change, and each finding has its own link to share with the team that owns it.
A service that comes with Observe
Corbado reads your error data, groups the errors that share a cause, measures what they cost and writes the finding. After a fix ships, Corbado measures again and marks it verified or regressed.
To explore errors yourself, use the Passkey Errors and User Debugging views. To hear about a problem as it starts, set up Alerting.
How findings are measured and used
See how Observe turns your authentication errors into root causes, sized in lost logins and ready to hand over.