Please Hold Off

What a country CEO actually gets from the group's AI program, and what the wait costs.

Illustration: a long patient queue of business leaders holding paper tickets waits at a corporate counter, where a clerk tends an enormous ledger.

I sell AI transformation to CEOs, and many of them run a country or a region inside something much bigger. Fifteen hundred people in the Nordics, inside a group of fourteen thousand, is a normal case.

They sit at the top of their own organization and in the middle of someone else's. They answer for a full P&L, inside a corporate structure they did not design. Most of the time that works. If they want to reorganize sales or rethink pricing, it is largely their call.

AI falls into a different box. Because it looks like technology, it lands on the desk of group IT. A decision about how the local business runs, paid for out of the local P&L, suddenly needs someone else's approval.

This fall, when these CEOs take our proposal to group, they keep getting the same answer. It is polite, and it comes quickly.

Please hold off. We are about to launch an AI enablement initiative from corporate.

Illustration: a fast ship with every sail set strains against a taut gold rope tied to an enormous flagship still being built in dry dock, where an officer politely raises a small flag. Far out at sea, a rival sail is already heading away.

It is a reasonable thing to say, and most of them accept it. Nobody wants to run a side project against their own group. But if you are one of those CEOs, it raises a question worth asking first. What exactly is that initiative, and what will my country get out of it?

What is coming from group

Over the past year I have worked with the leadership teams of more than a dozen European companies. Group AI programs almost never look different from each other. Either IT runs it, and AI soon becomes a question of tools. Or a central team runs it, clever people with far more countries than hours.

Either way, you can expect a platform and licenses, along with security, privacy and cost control. You can expect governance, and safe access to the ERP, the CRM and the finance systems. There will be shared company context for every agent to draw on. And there will be a handful of processes that really are the same everywhere, such as the monthly close and the quarterly business review.

All of that belongs at group. You do not want to build any of it yourself, and twenty countries buying twenty tools would be a mess.

What it will not include, and I doubt anyone at group would claim otherwise, is the twenty or thirty processes that actually run your business in your market. Nor the change management that goes with them. Nor your leadership team's own journey, from using AI to leading differently because of it.

Diagram: a building in cross-section. Below the ground line, group's foundation: platform and licenses, security, privacy and cost control, governance, data access, company context, a few global processes. Above it, what stays with the country: the local processes, change management, the leadership team's journey, testing on real customers.

Then there is timing. Countries queue for onboarding, and somewhere at head office there is a very good spreadsheet of who is next. In the companies I work with, nine to twelve months is normal before the first real working session on a local process.

So what is on the table is the foundation, in about a year. What you need to know is what that foundation does for your numbers on its own.

What tools do on their own

You already have an early answer. Many of your people use AI today, and some will tell you that parts of their work now go three times as fast. That is real. But if all you have added so far is tools, ask your CFO where it shows up in the P&L, and expect a pause.

To see why, follow one proposal through your company. Your version will have more steps and different names. The shape will be the same.

A salesperson sits down on Monday morning to write it. It used to take her most of the day. With AI it takes an hour, and it is better. That afternoon it goes to her manager for price approval. He is traveling. On Wednesday he has a question. Legal wants to see one clause. On Thursday it reaches the customer.

Before AI, it also reached the customer on Thursday.

Diagram: one proposal across Monday to Thursday. With AI it is written in an hour instead of most of a day, and it still reaches the customer on Thursday.

That is individual productivity. The gain is real, and it stops at the edge of her desk. More licenses give you more of it. Data access makes the answers better, and you will need it later. Neither touches Thursday.

Illustration: a long iron pipe with four leaks: only a slice of the job, the next step queues, someone has to check it, freed time goes nowhere. A trickle reaches a gold bucket labelled the P&L.

What it takes to change Thursday

Thursday is a decision about who approves what, usually written after a deal that cost someone margin. Change it, and the picture changes. Routine quotes go straight out once an agent has checked them against your pricing, and only the exceptions reach a manager. The proposal leaves on Monday afternoon.

Diagram: with the approval rule changed, the proposal reaches the customer on Monday, three days earlier, and only the exceptions reach a manager.

It is the same AI. A step stopped, and that is what turned a personal gain into a business gain. This is process redesign, and it is where I would look for your money. It is also where the three times your people talk about starts to reach the business.

Consider who can make that call. Sales owns Monday. The sales manager owns the approval, legal the clause, finance the margin. None of them can remove the wait, because each owns only a piece of it. It takes someone above all of them, who also knows which old rules still protect margin and which are just habit. In a country, that is you and your leadership team. Group cannot get that view from a distance.

Once the quote goes out within the hour, a further question opens up. On some deals, that speed may be worth paying for. You have then stopped changing a process and started changing what you sell. That is the third place a gain can land, business model and new customer value, and it gets the best slides.

A corporate strategy can describe a new business model. It cannot find out whether anyone will pay, because the customers are not at group. The decision to change what the whole group sells belongs to its CEO and owners, as it should. But until a country has tested it on a real customer at a real price, that decision rests on theory.

Keep this in proportion. Most of your money will sit in process redesign for a while yet. But one customer for one quarter costs little, and it produces the only real evidence your group is going to get.

So there are three places a gain can land, and three questions that find them: who got faster, what stopped, who pays differently. An IT rollout with tools and data access serves the first and prepares the ground for the second. It changes no process and no offer.

Diagram: three places a gain can land. Individual productivity: who got faster? Team and process redesign: what stopped? Business model and new customer value: who pays differently?

Waiting for it therefore does not buy you the effect. It moves the start of the real work back by however long the wait turns out to be. In my experience, that is about a year.

Meanwhile, your competitors

A year would cost little if the market waited too.

I do not mean the competitors on stage. You will not find the real change at a capital markets day, because that is where the AI slides are. The real change does not announce itself. It shows up as a market that has become harder to deal with. A rival comes back with a firm price while your proposal waits for approval. Tender responses improve from a team that has not grown. Deals start taking longer for you and not for them.

Your people will not report that a competitor has redrawn its quoting process with AI. They will tell you the market has gotten tougher, and they will be right.

Illustration: an executive smiles as she reads a letter with a gold seal, while a breathless young man arrives in the doorway with his own envelope, one beat too late.

So you cannot really wait, and you have no interest in a fight with your own group. There is a third option, and group may like it more than you expect.

Going first, on group's terms

My advice is to say yes to group, and mean it. You want what they are building. Then ask to be the country that goes first, alongside it. Use what group has already approved, keep customer and employee data out until group has cleared it, and hand over everything you learn.

Start with your own leadership team, because nobody redraws a process around a tool they have never really used. Then pick the one process that earns or protects the most money in the country, and make it the leadership team's problem rather than IT's. You do not need a business case to begin. You do need a way to be wrong: one number that should move, one that must not get worse, and a date to look at both.

I have watched this work. One industrial group I work with began with one country's sales team. What they built is where the next market now starts, and it led to an agreement at group level.

Put one question to group as well. Which of my processes will work differently a year from now because of your program, and who will redraw them? If there is a real answer, wait and help. More often the honest answer is that this is country work, and then you have both agreed on what is yours.

Group needs this more than it can say today, because a platform needs something to run on it. A few weeks after launch, the email everyone saw coming will arrive: could each country please share its most promising use cases? "We waited, as you asked" is a true answer. "We took three days out of our quotes, and here is how" is a better one.

When the platform arrives, will it still be Thursday in your building? And what day will it be at your competitor's?