Metrics

Failure risk cost

Failure risk cost is the expected cost of future breaks: this quarter's finished changes times the fitted likelihood of a break times the mean cost of one break, per change class. It is assumed even when nothing broke, because the risk is in the change's shape, not in luck.

The likelihood is per change class (how far the change reached across interfaces, and how tested those boundaries are). Until it is fitted on real incidents it uses published starting values, and the report marks them as starting values, not fitted. The cost of one break is the mean realized engineering hours of breaks whose causing change is in that class. Business impact is not counted.

Back to Help

Measure the overspending. Then change what made it expensive.

Defusor is in its testing phase: free for a few Series B or C SaaS companies, in exchange for feedback.

Request a free test