The method grew out of four failed AI transformations.
Not out of a new theory of work, but a simple inverse argument. If resistance is the main reason AI programs fail, then it helps to measure resistance before procuring the technology.
How it came about
Four AI transformations ran the same way. The first time, we sat on the steering committee and actually thought the idea was good. Map processes, unlock potential, calculate business cases, the grown-up version of digitalization. Twelve months later, the slides sat on the drive, and in the department they were intended for, no one used the system. At the time we blamed it on change-management mistakes.
The second time, we paid better attention and produced the same result anyway. The third time was no different. By the fourth time, it dawned on us. Maybe it wasn’t the change management, but the starting point.
Then the operations manager of a mid-sized company (300 employees) called, asking for an AI strategy. This time, no process map. In the first week we talked with twenty people. No agenda, no assessment form. And in those twenty conversations, the method was born.
Silke Rengstorf
Senior Coach & Management Consultant
Coaching and consulting in transformation processes. Runs the individual conversations, facilitates the heatmap workshops with the teams and supports managers during AI adoption.
Michael Kupermann
Management consultant for IT & automation, machine learning & AI
Framework and architecture of the FireScore. Owns the survey, the evaluation and the feasibility check of the automations with IT.
The conversation that brought forth the FireScore
On the third day we sat in a meeting room. A back-office clerk in the morning, a salesperson in the afternoon. The same list, the same stop at “monthly reporting”, the same outputs. And two completely different answers.
Those are the worst two days of the month. Numbers from SAP into Excel. Formatting. Checking rounding. And in the end no one really looks at it properly anyway.
That’s my window in the month when I really understand the company. Everything else is day-to-day business. Those two days are my thinking time.
By classical logic this was a clear automation case, repetitive data extraction at high volume. We would have freed one person and taken something away from the other, who would not have protested openly. The objection would have arrived as a question to the Works Council about whether such a system was compliant with data protection law at all, and later as edge cases that genuinely turned up in testing. Around 70 percent of it would have been right.
The right question is not what a machine can do, but what people are willing to entrust to it and hand over.
The next morning, the four dimensions were written on a sheet of paper.