It’s always the same story: the AI budget is approved, management is on board, everyone’s enthusiastic—but then comes the question that opens Pandora’s box: “Which process should we start with?”

And that’s when something predictable happens: opinions start flying. Operations says billing, because “that’s where we waste time.” HR says onboarding, because “that’s where you see the value.” IT says ticketing, because “at least we have the data.” Everyone is right from their own perspective, but no one has a common standard.

We know the outcome: three pilot projects launched simultaneously, the budget spread thin, and none of them made it to production. The problem wasn't the technology, the context, or the execution—it was prioritization.

A method, not an opinion

The Process Priority Assessment (PPA) exists to answer that question in a justifiable way. The logic is simple: identify candidate processes, jointly define the evaluation criteria (the “drivers”—such as effort, costs, and customer value), weigh them collectively, and evaluate each process against each driver. The result is a ranking that each function has helped to create and that is presented to the steering committee, with a clear understanding of the basis for each figure.

An interesting detail: the scale isn’t from 1 to 5; it’s a choice between 1, 3, or 9. This isn’t just a workshop gimmick—it’s the classic relationship scale from Quality Function Deployment, the “House of Quality” introduced to the West by Hauser and Clausing in the *Harvard Business Review* in 1988. Half a century of industrial prioritization has already established that the average score is the real enemy. To give a 9, you must be convinced that the process is truly a priority based on that criterion; to give a 1, you must be convinced of the opposite. The discrepancies that emerge map out the areas where the organization is not aligned. These are points to discuss before investing, not after.

With one caveat: the PPA does not eliminate politics. Whether the drivers are chosen by a single person, or whether the weightings are determined after observing the race proceedings, the method produces flawless numbers that serve the same decision-making process as always—with the added appearance of objectivity. Power does not disappear; it shifts upstream, to the criteria, where at least it is visible and open to debate.

Why AI Is a Game-Changer

PPA predates generative AI. Effort in FTEs, repetitiveness, and degree of standardization: Lean first, and then rule-based automation (e.g., RPA), were already addressing these questions. What has changed today is not the list of drivers, but their relative importance and the scope of what can be considered for automation.

First, the data. Rule-based automation didn’t need a clean data history; all it needed was a stable interface. AI is different. A costly, repetitive process that everyone feels the impact of—but with data scattered across three systems that have never been reconciled—it’s not a viable candidate today: it’s a data governance project in disguise. Data quality ceases to be a technical detail and becomes a driver that, on its own, can rule a project out.

Second, the scope is shifting. Lean and automation targeted the structured aspects of work and left out everything that required judgment, documentation, or language. Today, that area is the new frontier: processes that had already been streamlined by Lean are back on the list, and processes that were never included because they couldn’t be standardized are now being added.

Third, a factor that didn’t exist before: error tolerance. An RPA bot makes deterministic and visible mistakes, while a generative system makes plausible mistakes. A process must therefore also be evaluated based on the cost of an error that goes unnoticed for three months—a criterion that was missing from traditional prioritization matrices.

The Next Step

An effective PPA is one you can explain and defend. And that is exactly where the AI Super Power master’s program at H-FARM Business School concludes: after building expertise in data, agents, and AI governance, the program culminates in the PPA—the moment when everything you’ve learned up to that point becomes a decision to implement within the company. Which process, and why that one.

The same question will come up again and again: with every budget cycle, with every new wave of technology. What changes is what you bring to that discussion: an opinion to defend, or a criterion that all departments have helped to establish—and that no one can dismiss with a simple “I don’t think so.”

The difference isn’t apparent on the day of the meeting. It will become clear six months later, between two companies that today seem to be in the same position: on one side, three pilot projects stalled halfway through; on the other, a process already in production, the next one already identified, and a reasonable explanation for why that particular one was chosen.

Lorenzo Ava