Quick answer
Anti-money-laundering rules oblige every business that serves you, from the company service provider and the notary to the bank, to build and keep its own file on you. The law lets them rely on each other’s work, but it also leaves the liability with whoever relies, so in practice each one starts from zero and one client ends up in three or four separate dossiers. Each dossier costs staff and software, and the client pays for all of them. A small part of that paperwork reduces risk. Most of it exists so that the organisation holding it can show an inspector that it asked.
I run compliance for a living. Eesti Firma is a licensed trust and company service provider, which makes us an obliged entity: one of the businesses that anti-money-laundering (AML) law, applied in much the same form across the EU, requires to check its customers and report what looks suspicious. Every client who registers a company through us is identified, screened and asked where the money comes from, and I believe in doing that properly. Identifying a straightforward founder and screening him against the sanctions and politically exposed persons lists takes a trained person about 15 to 20 minutes. The rest of the onboarding is paperwork built around that quarter of an hour, and the same paperwork is then rebuilt, from the beginning, by every other obliged entity the client meets.
This is not an argument against know-your-customer (KYC) checks. It is an argument about what they have turned into. Everyone in the chain does the work carefully, because everyone fears the consequences of not doing it, and somewhere along the way that fear became the point. I have come to think that the file exists less to stop money laundering than to prove, when the inspector comes, that the organisation holding it asked the right questions.
One Client, Three or Four Rounds of Checks
Here is what the first few months look like for a typical client of ours, a founder from Germany, Spain or Turkey who wants an Estonian private limited company and a working account. The sequence would be much the same in Dublin or Lisbon, because the rules come from the same EU directives; Estonia adds a step of its own for e-residents.
- The state, for e-residents. A founder who wants Estonia’s digital ID applies for e-Residency, and the Police and Border Guard Board runs a background check before the card is issued. This is not an AML check in the legal sense, it is vetting under a different act, but it asks for the same passport and the same explanation of purpose, and the applicant pays a state fee for it.
- The service provider. We verify the founder’s identity, establish the beneficial owner, screen him against sanctions and PEP lists (a PEP is a politically exposed person, someone whose public office makes bribery a risk), ask what the business will do and where the start-up money comes from, and assign a risk level. That is a questionnaire, a set of documents and usually a video call.
- The notary. A founder without an Estonian digital ID sets the company up through a notary instead, and the notary is a separate obliged entity with the same duties. Our file carries no weight in their office; they identify the client and check the owner themselves.
- The bank or payment institution. Whoever opens the account, a bank or a fintech, runs its own form, its own document list and its own interview: the business plan, the expected volumes, the main counterparties and the source of funds, again. A second account, which many of our clients need, means the same round once more.
Depending on the route, a founder goes through three or four of these, and the checking does not stop there. When the client pays our invoice, our own bank may ask us who the payer is and what the money is for. An accountant who later takes the company on is an obliged entity too. And each of these organisations must keep its file for at least five years after the relationship ends, keep it current, protect it as personal data and monitor against it, because each is separately liable for what it holds.
Every round is paid for separately. A provider like ours employs people to read questionnaires, chase documents, screen names, write up a risk assessment and then maintain the file for years, and that work is covered by the fee. The notary bills time. Banks are the biggest spenders by far; one industry survey put the cost of financial-crime compliance across the global financial sector at over $270 billion in 2022. Nobody adds these numbers up, because each sits in a different budget, but the client pays all of them, in fees and in state fees, and the compliance staff at each stop never see what the previous stop found. And like the rest of the EU rulebook, it is a fixed cost: a bank absorbs it, a provider with a dozen staff feels it in full.
How a Form Became a Dossier
Ten years ago the checks existed, but they were a form, not a dossier. Opening an account meant a passport, a signature and a few questions, and a provider like ours kept a thin file. Then came the scandals. The EU rewrote its anti-money-laundering directive in 2015 and again in 2018, each time adding obliged entities and obligations. Estonia got a new national act in 2017, and in 2018 the Danske Bank report put a figure of around €200 billion on the non-resident money that had passed through one Tallinn branch, much of it later judged suspicious. A small country became a byword for laundering.
Danske was not a shortage of forms. The bank had files on its non-resident clients and ignored what its own monitoring showed, and the supervisors did not look hard enough at one institution. The response, all the same, was more forms for everyone. Every bank and every supervisor did what any organisation does after public humiliation: it asked more, kept more and trusted nobody. In my experience the cycle only runs one way. A supervisor is blamed for the scandal it failed to prevent, not for the questionnaire it made ten thousand honest clients fill in. An obliged entity is fined for the field it left empty, not for the field it added. Simplification programmes exist, in Brussels and in Tallinn, but nobody’s inspection depends on them, so questions accumulate faster than they are removed.
The Law Allows Sharing. In Practice Nobody Shares
Identity, to be fair, is shared already. When an e-resident or an Estonian resident logs in with a digital ID, every bank and provider is relying on a state-verified identity. What is not shared is everything else in the file: who the owner is, what the business does, where the money comes from. The law permits that too. Both the current EU directive and the new EU Anti-Money Laundering Regulation, which applies from July 2027, let an obliged entity rely on customer checks already performed by another obliged entity, and the new EU authority AMLA is to issue guidelines on how. So why does the bank ask our client to start from zero?
Because the responsibility stays with the one who relies. If our file turns out to be wrong, the fine goes to the bank, and the bank’s supervisor will ask the bank, not us, why it trusted a firm it does not control. Assessing another company’s procedures costs more than sending the client a fresh form, so every institution makes the same rational choice. The law permits reliance on paper and punishes it in practice, and the sum of all those sensible individual choices is the theatre of the title. The file has stopped being a tool and become a product: its job is to exist, to be complete and to be producible on demand, so that whoever holds it is covered.
Company formation is my example because it is my work, but the pattern runs through the economy. A flat sold in any EU city produces a file at the estate agent, another at the notary, a third at the lending bank and sometimes a fourth at the bank transferring the buyer’s savings, four files on the same two people for one transaction, with none of them drawing on the previous one. Change more than €1,000 in cash at an Estonian currency exchange and you will be identified and recorded, and the exchange will keep that record for years next to thousands like it. Accountants, auditors, lawyers and dealers in valuables hold files of the same kind.
What in the File Reduces Risk
The standard file that banks and providers across the market collect from a new company falls, as I read it, into three groups: the part that makes it harder to misuse a company, the part that exists mainly so the holder can satisfy an inspector, and the part that is pure ritual. The table is my own classification, not a legal one, and the test in each row is a simple one: does anyone ever act on this piece of information?
| Element of the standard file | What it does |
|---|---|
| Identity verified against a document, with a live or video check | Reduces risk. A real person is attached to the company. |
| Beneficial owner established and checked against the register | Reduces risk. Hidden ownership is what a launderer needs most. |
| Source-of-funds documents: contracts, statements, sale deeds | Reduces risk, if someone reads them. A signed declaration that the funds are legal adds nothing to the documents. |
| A plain description of the business: who pays whom, for what | Reduces risk where the holder monitors transactions against it. A bank does. A notary or a formation agent never sees a transaction, yet collects the same description. |
| Sanctions and PEP screening | Reduces risk, and is cheap, because software does it. |
| Expected annual turnover, to the nearest thousand, for a company that does not yet exist | Useful to a bank as a baseline for monitoring. For every other holder of the file it is a number nobody will look at again. |
| Self-declaration that the client is not a PEP and acts on his own behalf | Mainly for the inspector. It shifts blame to the client if he lied; it does not detect anything. |
| The same set collected again by each institution | For the inspector, as explained above. Each copy protects its holder; the second copy does not protect the system. |
| Periodic re-KYC, the full review every one or two years, when nothing has changed | For the inspector. It proves a calendar was followed. |
Two things follow from the table. First, the file earns its keep only where someone monitors against it afterwards, which in the chain above means the bank, and the bank alone. Second, length comes from the middle group. An inspector cannot see whether we understood the client, but can see whether field 14 was filled in, so fields are added and rarely removed.
What the Arithmetic Says
I use Estonian figures because they are the ones I know best and because the Financial Intelligence Unit publishes a yearbook. In 2024 the unit received 14,185 reports from obliged entities. It made 133 analysed transmissions to investigators and about 1,700 further transfers of information, and investigators seized €5.7 million of assets the unit had frozen. According to the national risk assessment, the unit’s analysed transmissions led to ten new criminal proceedings in 2024, and over the five years to 2024 Estonian courts handed down 32 judgments in money-laundering cases.
Now the other side. More than 25,000 new private limited companies are registered in Estonia each year, the register holds around 370,000 legal entities of all kinds, and the e-Residency programme counts over 134,000 e-residents. Each new company, each account and each digital ID means at least one full KYC file, usually several. My rough estimate, and I stress that it is mine, is that obliged entities in Estonia produce well over a hundred thousand KYC files a year, and that the whole system turns them into ten criminal proceedings and a few million euros seized. Suppose one client in ten thousand has criminal intent. The other 9,999 go through three or four rounds of checks so that he can be found, and the figures above say that he usually is not.
There are two fair objections, and I want to state them. One is deterrence: criminals who never apply because they know they will be asked do not appear in any statistic. That is true, and it is also why nobody has ever had to prove it. The other is scale: one laundering scheme can move more money than a thousand honest companies earn, and if the files had stopped Danske they would have paid for themselves many times over. But the files did not stop Danske, and the convictions the FIU describes in its yearbook began with reports of suspicious transactions, not with questionnaires. The onboarding file is where the money is spent; monitoring is where the results come from. The international evidence points the same way. Europol estimated in 2016 that only about 1% of criminal proceeds in the EU are ever confiscated, and the researcher Ronald Pol, who tried to put costs and recoveries side by side, concluded that compliance costs exceed recovered funds more than a hundred times over. No supervisor in Estonia has published a comparison of that kind for the system it runs.
I do not expect the balance to improve. The new EU rulebook adds a European supervisor, an EU-wide cap of €10,000 on cash payments and several new categories of obliged entities, from crowdfunding platforms to football clubs. Each new category means another drawer. Each future scandal will add fields, because adding fields is how an institution shows it has learned. And a single European rulebook makes any rollback harder, since a question written into a regulation cannot be removed by a national parliament that notices it is useless.
Why Nobody Says This Out Loud
It is not fashionable to weigh this. In public, the only safe position on money laundering is more: more checks, more reporting, more zeal. Anyone who asks whether the last round of rules achieved anything risks being heard as someone who wants fewer rules for criminals, so the compliance officers and lawyers who, in my experience, share many of these doubts in private keep them there. Fighting money laundering has become a cause more than a policy, and causes are rarely asked for their cost-benefit figures.
That is worth saying plainly, because of where the road leads if nobody weighs it. A Europe that has written the strictest privacy law in the world has, with the other hand, obliged millions of private businesses to keep dossiers on their customers for years, to report them to the state on suspicion, and to answer for it if the dossier is thin. Every one of those files was opened in good faith to stop a crime. Taken together they amount to a standing record of who owns what and who paid whom, held in drawers that answer first to an inspector. I do not think anyone designed that. I also do not think it should keep growing without someone asking what it is for.
What I Would Change
Most of this does not need a new law, only different decisions about how the existing one is applied. The first point does.
- One verified file, carried by the client. Someone verifies the person once and takes liability for that verification, backed by a licence, insurance and audit, so that the institution relying on it is not the one punished if it is wrong. That verifier could be a bank, a notary, a licensed provider or a state service built on the digital ID; what matters is where the liability sits. Today’s rules, national and EU alike, put it on the one who relies, and that is why nobody relies. Changing it needs legislators, and AMLA’s forthcoming reliance guidelines are the natural place to start the discussion.
- Supervise outcomes, not forms. An inspection should ask what a firm’s monitoring produced and where its reports went, not whether every field on every form was filled in.
- Forms that match the company. A new company with one founder and no turnover needs two pages, not twenty. The enhanced questions, the deeper checks the law reserves for higher-risk clients, belong at the point where activity starts and there is something to examine.
- Re-KYC on change, not on the calendar. A new owner, a new country or a new line of business is a reason to look again. The passing of two years is not.
- Use the data the state already has. For an existing company the Commercial Register holds the owners, the board and the annual reports. An obliged entity should pull them from the register instead of asking the client to retype them. Estonia’s Supreme Court applied the same logic to the supervisor itself in January 2026, when it ruled unlawful a 44-question form the Financial Intelligence Unit had sent to all licensed service providers, partly because the state should first use the data it already holds.
Getting Through the Checks Without the Theatre
At Eesti Firma we ask for what the law and our own risk assessment require, and we explain why each document matters. We also help clients build one set of documents and answers at the start that the bank and the payment institution can be given without rewriting, because in my experience what keeps an onboarding short is not a shorter questionnaire but a business described the same way to everyone who asks. For companies that become obliged entities themselves we draft the policies and procedures the law demands, which is, I admit, part of the same theatre; the difference we can make is to write them so that they are usable rather than merely complete. If you are setting up a company and want the compliance part handled by people who read what you send, we can help with company formation and with everything that follows it.