Every AI business case rests on a number almost nobody measures: what the process costs today. Not its salary line — its full price. The unmeasured process is unpriceable: the business case for automating it becomes an act of faith, the vendor's slides supply the arithmetic, and the company discovers the real cost of the process a year later, in the failure of the project that was supposed to remove it.
Pricing the process honestly changes the AI decision before any technology enters. First, it reveals where the money actually is: automation aimed at the visible time often misses the loops and the waiting where the real price sat — the robot automates the cheap part and the expensive part survives untouched. Second, it exposes the processes that should not exist: a surprising share of «automation candidates» are artifacts of an old constraint — approvals for exceptions that no longer happen, reports nobody opens — and the honest price of such a process is a deletion order, not a machine learning model. Third, it sets the baseline: the number the project will be judged against, agreed before the vendor's slides arrive and inflate it.
Half the automation business cases die not because the technology failed, but because the process was never worth what everyone assumed it cost.
The method is unheroic: follow the process for a week, count the cases, price each component at honest internal rates, and write the number where the business case can see it. This is the first step of use-case selection and the entry test of the P&L question; it is what keeps the portfolio built by an opportunity assessment honest, and what separates strategy from shopping. Companies that skip it are not choosing worse projects; they are choosing blind — and paying for the seeing later, at project prices.
We use cookies to analyse traffic and improve the site. Analytics is enabled only with your consent. Cookie Policy