Predicted Risks

Predicted Risks uses AI to identify key results that are likely to miss their target before they officially go off track. Instead of waiting until a KR is obviously behind, it analyses check-in velocity and projects where each KR will land by cycle end.

How it works

Every day, OKRs Tool calculates the velocity of each key result — how fast the value is moving based on recent check-ins — and projects it forward to the cycle end date. If the projected value will miss the target, the KR is flagged as a predicted risk.

Each risk is assigned a severity level:

• Critical — projected to miss target by more than 40%
• High — projected to miss target by 20–40%
• Medium — projected to miss target by less than 20%

Where to find predicted risks

Predicted risks appear in two places:

• Dashboard — a compact card shows the top 3 risks. Click "View all predicted risks" to go to the full page.
• Reports → Risks — the full risk report with severity filters, confidence scores, AI-extracted blockers from check-in comments, and suggested interventions.

Suggested interventions

For each flagged KR, OKRs Tool suggests up to 4 interventions:

• Nudge owner — send the KR owner a prompt to check in or unblock
• Adjust target — open the KR editor to recalibrate
• Add initiative — link a new action to this KR
• Schedule 1:1 — create a Google Calendar event with the owner

Confidence score

Each prediction includes a confidence score (0–100%). Higher confidence means more check-in history is available to base the projection on. A KR with only one check-in will have a lower confidence score than one with eight weeks of data.