Certified public accountant (KICPA) · Seventeen years in audit
I spent most of my career looking for the one account that had drifted away from what it should have been. ONKI applies the same question to people.
Financial statement fraud is rarely caught by an absolute threshold. A number is not suspicious because it is large. And a single point in time tells you very little — cut the year off at the right moment and almost any set of statements can be made to look healthy.
So auditors do two things instead. They look at movement rather than level, and they compare that movement against companies that had, until then, been moving in much the same way. The question is never is this number bad. It is why has this one pulled away from the others.
That logic transfers. Typing speed is not interesting on its own — being slow is not a problem, and plenty of people have always been slow. What is interesting is someone whose own rate of change has separated from the peers who were changing at the same pace.
That is the whole idea behind ONKI. It also explains why the tool makes no diagnosis: an auditor flags an anomaly for someone else to examine. The flag is not the verdict.
The second one is recent, and deliberate. Building a trial taught me quickly how much I did not know about how health interventions actually reach people — which is also why I am at BISHAC this year.