A group rarely loses control of itself in one decision. It loses control in a sequence of small ones, each defensible on the day it was taken. A country manager needs signing authority to open a bank account, so he is given it. A local partner will only commit if he holds shares in the company he runs, so he gets them. A third subsidiary is incorporated in a hurry and goes under the second, because that was the fastest route. Three years later the client asks a simple question, what do we own and who decides, and nobody in the room can answer it in a sentence.
For the adviser accompanying that expansion, this is the part of the work easiest to postpone and most expensive to repair. Nothing about it is urgent while it is still cheap to fix. By the time it is urgent, fixing it means reopening negotiations with people who now have leverage they did not have at the start.
Erosion happens between the entities, not inside them
Most advisers reviewing a growing group look at the entities one at a time, and each one is usually fine. Clean articles in Spain, a properly appointed managing director in Germany, a competent notary in the Netherlands. The problem is rarely inside a single entity. It is in the relationships between them, and no file contains those.
What accumulates is a gap between the ownership chart and the practical answer to who decides. Ownership says the parent holds everything. Practice says the country manager sets prices, signs the leases, hires and fires, holds the customer relationships in his own name and controls the only bank mandate anyone uses. The parent can outvote him at a shareholders meeting, a remedy nobody wants to use, because the day it is used the business in that country stops working.
That gap is invisible in a corporate search and obvious in a due diligence interview. Buyers and lenders read the chart, then talk to the people, then price the difference.
The questions worth answering before the first entity exists
The most useful conversation an adviser can have happens before any foreign company is formed, and it is not a legal conversation. It is a set of plain questions about how the group intends to work. Which decisions belong to the parent and which are local. Who owns the customer contracts. Where the intellectual property sits, and whether the local entities are licensees or owners. Whether the local business is meant to be sold separately one day.
Clients often answer quickly and inconsistently, which is itself the finding. A founder who says the parent decides everything and then describes a country manager with a free hand and a profit share has not thought it through. That becomes a problem once it is written into three sets of articles in three countries.
The output is short. A single page setting out which matters are reserved to the parent, what a local entity may do without asking, and what must never sit outside the parent. Everything drafted afterwards, in any jurisdiction, is measured against that page. Where local law will not carry what the page requires, the adviser learns it early, while the structure can still change.
When the local partner wants equity
The request is reasonable and it usually arrives at the worst moment, when the client has committed to the market and the partner knows it. The instinct is to concede shares in the local operating company because that feels contained. It is not. A minority holder has statutory rights in most European jurisdictions that no commercial agreement fully displaces: rights to information, to challenge resolutions and, in some cases, to block corporate actions the group will eventually need.
There are usually better instruments for the same commercial purpose. A profit participation tracking the local result gives the partner the economics without the governance. An option exercisable only into the parent aligns him with the group rather than his territory. Where local equity is genuinely the only answer, the terms that matter are the ones nobody wants to raise at the start: what happens when he leaves, how the stake is valued, and whether the group can buy him out without his consent.
The other point worth insisting on is that a local partner holds shares in the local company and nothing else. Partners who end up with a stake in an intermediate holding company, because it was convenient at the time, acquire a veto over every other country in the group. That is usually the residue of a hurried incorporation.
Signing authority leaks faster than shares
Shares are negotiated. Authority is handed over in an email. Somebody has to sign a lease while the directors are travelling, so a broad power of attorney is issued and never withdrawn. A bank asks for an authorized signatory and the local manager is the only person in the country, so he becomes the sole mandate holder. Within a few years the parent can own everything and sign for almost nothing without local cooperation.
Powers of attorney should be specific, limited in time, registered where the jurisdiction requires it, and reviewed on a fixed date rather than when something goes wrong. Bank mandates should require a second signature above a modest threshold, held at parent level, and the parent should have read access to every account from the day it is opened. None of this is expensive. It is simply never anyone’s job.
The same discipline applies to what gets overlooked because it is not a corporate document. Domain names, administrator accounts, the customer database, the licence to the software the business runs on. A client can hold every share in a subsidiary and still be unable to operate it because an account is in a departed manager’s personal name.
Where the board actually meets
Formal board composition is the part most advisers get right. The parent appoints the directors, the appointment is registered, the file is clean. What matters more is where decisions are actually taken and recorded, because that determines both control and, separately, where the group is treated as managed for tax purposes.
A group that expands quickly tends to develop a real decision making body that appears nowhere in the corporate records. A weekly call between the founder and the country managers, or a chat group. Decisions taken there are implemented by whoever has authority locally, and the minutes, if they exist, are written afterwards to match. That works until someone must prove what was decided and by whom.
Bringing that body inside the structure is usually straightforward. The same people meet on the same rhythm, but the meeting is constituted, its decisions recorded, and matters reserved to the parent visibly taken by the parent. In a Dutch holding, the articles can be drafted so that defined categories of decision require approval from the general meeting or a supervisory body, which turns a habit into an enforceable right. What cannot be repaired retroactively is a long record of decisions taken in one country and papered in another.
Cross holdings and the structure nobody can unwind
Speed produces strange charts. A subsidiary is incorporated under whichever entity was already open. A second country is added below the first because the first had cash. An intermediate company is created for one transaction and never removed. Each step was rational on its own. The result is a group where three countries hang off an operating company and one entity holds shares in something above it.
The cost is not theoretical. A group that wants to sell one country discovers it cannot separate it without a chain of consents and a tax event nobody modelled. A lender asked to finance at parent level finds the assets three levels down behind minority holders. A family bringing in the next generation finds the shares carrying the value are not the shares the parent holds.
Untangling a structure is always possible and never as cheap as building it correctly. The working rule during an expansion is that new entities hang directly off the holding company unless there is a specific written reason otherwise, and that any intermediate company created for a transaction has an end date. Reviewing the chart once a year against that rule takes an hour.
What belongs in the articles and what belongs in the agreement
Advisers from common law jurisdictions tend to put everything in the shareholders agreement, because that is where it goes at home. In continental Europe that instinct leaves value on the table. An agreement binds the people who signed it and gives a claim in damages when breached. Provisions in the articles bind the company itself, are visible to third parties, and can make a breaching act ineffective rather than merely actionable.
The distinction is practical. A transfer restriction in the articles can stop a share transfer from taking effect. The same restriction in a private agreement gives the other shareholders a lawsuit after the transfer has happened. Dutch law allows considerable freedom in drafting the articles of a BV, including share classes with different rights, restrictions on transfer, appointment rights attached to a class, and instruction rights in favour of the general meeting. Much of the control architecture the client wants can live there.
What stays in the agreement is what should not be public and what is specific to the current shareholders: the economics between them, undertakings that will lapse, the arrangements for exit. Which layer carries which provision is worth deciding deliberately in each country, with counsel who knows what will actually be enforced there, rather than by translating a template.
The test is the next transaction
Every structure is eventually read by somebody who is not paid to be sympathetic: a buyer’s counsel, a credit committee, a tax authority. They ask the same three questions in different words. Who owns this, who controls it, and can the ownership be transferred without asking anyone’s permission. A group that answers all three quickly is worth more than an identical business that cannot, and the difference shows up in price and conditions rather than in a refusal.
That is the argument to put to a client who thinks this can wait until things settle down. It is not a compliance exercise, and it is not about anticipating a dispute with people he currently trusts. It is about keeping the group in a state where the next transaction is a negotiation about value rather than about consents.
For the adviser leading the relationship, holding the group view is the work, and it does not require covering every jurisdiction personally. It requires a clear statement of what the parent must retain, and local counsel who drafts to it rather than to a local default. The Dutch layer is a common place for that discipline to break down, because the holding company is often incorporated fastest and reviewed last. It repays being treated as the part of the structure everything else is measured against.