Start from Outcomes, Not Headcount: Lessons from the Phison CEO on Not Cutting People with AI

Will AI reduce headcount, or not? This question comes up constantly, both in what executives say and in the news.

Let me state my conclusion first: I think focusing on the number of people is itself the mistake. In most cases it is employees who worry about headcount, and management responds with messages such as "we will cut" or "we will not cut." But what management actually looks at is outcomes.

What the Phison CEO said, and what reports confirm

Pua Khein-Seng, co-founder and CEO of Taiwan's Phison Electronics, has said in effect that cutting people because of AI is a foolish approach (reported by TVBS and others). Because the wording is strong, that line tends to spread on its own.

Here is what I could confirm in reports. According to United Daily News coverage of an October 14 talk, internal AI usage exceeded 80 percent in September, and the company aims to add output equivalent to 500 to 700 engineers by the end of the year. One example given is log analysis for verification, which went from three hours to 30 minutes. Reports also say that the AI rollout is led from the top. A Central News Agency article in June described a dedicated AI team and a policy of moving people to higher-value work rather than cutting them.

What interests me most is how the goal is set. "Adding the output of 500 to 700 engineers" is not about how many people to cut or keep; it is about how much to increase outcomes. The words are addressed to employees; the numbers are what management pursues. Read that way, his remarks and my own view overlap substantially.

The moment you focus on headcount, musical chairs begins

Saying that focusing on headcount is a mistake may sound as if I do not value people. I mean the opposite. When a discussion starts from headcount, people are the ones who get hurt most.

The moment a debate about "how many can we cut" begins, employees start wondering whether they will be among those cut. People start playing musical chairs as soon as they focus on numbers. Colleagues stop looking like partners and start looking like rivals for a chair. Then nobody tries to use AI well, because the better they use it, the more their own chair seems at risk.

Reform quietly stalls, and top performers leave first

A common reaction in a company playing musical chairs is defensive foot-dragging: not opposing openly, but not cooperating either. If top performers give up and leave while reform is stalled, one's own chair is safe. Even if no one says so, I believe many people inside companies quietly hope for that.

When the discussion starts with headcount, top performers are the most likely to leave. They are the people who want results, and they are in demand elsewhere. Those who play defense stay, and those who want to move forward go. A company that started from headcount ends up losing its most important people.

I do think people may change during a transformation. But that happens as a result of pursuing outcomes; it is a different matter from starting with a headcount target.

Grow the outcomes you have, fix the ones that are not coming

So what should you focus on? Very simply: grow the outcomes you already have, and fix the areas where outcomes are not appearing. Whether to introduce AI and how many people it takes come after that.

For example, if faster replies to inquiries lead to more orders, make those replies faster and more accurate. If slow quotations are costing you deals, fix that. When you start from outcomes, AI appears naturally as a tool for them, and it does not become a story about taking someone's chair. In my experience, there have been many cases where digital transformation moved more easily when we focused here, because a clear purpose makes it easier for people on the ground to cooperate.

The president decides first where the time freed by AI goes

If a three-hour task becomes 30 minutes, two and a half hours are freed. If you proceed without deciding where that time goes, it either gets absorbed by small tasks and disappears, or it gets pulled back into a headcount discussion: "If time is freed, aren't people left over?" What was freed is time, not people, and that time simply has no destination yet.

The criterion for deciding is for the president to make clear what they want the company to become and what outcomes they want. If you want more new customers, put the time into proposals and sales preparation. If you want higher satisfaction among current customers, put it into follow-up and consultations. If you want new products, put it into development and prototyping. Without a clear outcome in mind, this choice cannot be made.

Once decided, check whether outcomes moved

After deciding, check once a month. Is the time really going where you decided, and are outcomes moving as a result? Look not at how many people there are, but at what has grown and what has been fixed. One or two check items per destination are enough.

When you explain this to employees, I think it is better not to start with headcount either. "This company aims for these outcomes. To get there, I want this task to be faster. Please use the freed time here." In that order, people's attention turns toward outcomes rather than their own chair. Conversely, if the headline of your first report is "AI freed up the work of so many people," the floor will immediately start counting chairs.

What you can do starting tomorrow

Set aside the musical-chairs mindset and focus first on maximizing outcomes. That is the main message of this piece.

Concretely, write down one outcome you want to grow and one you want to fix in your company. Then, if AI has made some task faster, decide which of the two the freed time goes to, and check at the end of the month. The headcount discussion can wait until after that.

▶ Watch the full video (Japanese)
https://www.youtube.com/watch?v=C6Yb-cjwjXk