Your best customers have the most complex KYB. They should do the least work.

One European FX payment institution summed up its onboarding in an RFI sent to KYB vendors: "There is no dedicated client portal for onboarding or periodic reviews. All interactions with clients are handled manually by email." Its simplest clients probably don't mind. Its best clients do.

The clients who bring in most of the profit are rarely a single domestic company with one director. They are holding structures, groups spread across several countries, and regulated institutions with several layers of ownership. What makes their KYB complex (cross-border flows, layered ownership, large volumes, politically exposed persons) is often what makes them valuable. Yet they get the longest checklists.

Onboarding a simple company from register data in minutes is a solved problem, or should be. The real test is the complex file. That is what white glove onboarding is for: the institution does the work, so the client doesn't have to.

Key takeaways

• Value is concentrated, and it is rarely simple. In a typical large bank's corporate book, economic profit comes from roughly 20% of clients (Bain & Company). Those relationships tend to carry the cross-border flows, layered ownership and volumes that make KYB complex.
• The happy path is table stakes. A domestic entity with complete register data should be decided automatically. If it isn't, fix that first.
• Complex files fall back to email. When automation hits a second jurisdiction or a holding layer, most processes go back to manual checklists. Corporate clients complain of "limited use of existing data" and "piecemeal requests for documentation" (Bain & Company).
• White glove means asking only for the net new, even for the most complex structure. Entities, directors, ownership chains and documents are retrieved, reused or derived first. The client confirms them and supplies only what nobody else holds.
• Blended averages hide the problem. Setting aside "only" the complex files can mean setting aside a far larger share of revenue. Measure each complexity tier separately.
• Asking less is not checking less. Registry data comes with a source and a timestamp, and is cross-checked against what the client declares. Data the client types does not.

Why do your best customers have the most complex KYB?

Know Your Business (KYB) is the due diligence a regulated institution performs on a legal entity: identifying it, its legal representatives and its Ultimate Beneficial Owners (UBOs), screening them for Anti-Money Laundering (AML) risk, and rating the relationship.

For a sole trader, this takes minutes. For the clients that matter most, it does not:

• A private banking client may hold their wealth through a personal holding company that owns a property company and a minority stake in a foreign investment vehicle.
• An institutional investor in a fund may be a pension scheme, an insurer or a fund of funds, each with its own regulated status and governance.
• A corporate buying FX hedging may be a group with subsidiaries in six countries, each needing to be identified and linked to the same UBOs.

The richer the client, the more entities, jurisdictions and ownership layers you have to work through. This is not a defect in the client base. It is what high-value clients look like.

The numbers point the same way. Bain's analysis of a typical large multinational bank's corporate book found that economic profit derives from roughly 20% of clients, and that about 1% of clients account for most of the value (Bain & Company, 2017). Those large relationships are the ones most likely to involve cross-border activity, multi-layered structures and large volumes: the same factors that push a file into enhanced review.

Risk and value are not the same thing. Some high-risk files are small: shell companies, entities in high-risk jurisdictions, thinly capitalised start-ups. So before calling the job done after automating the standard journey, run one simple split on your own portfolio: by risk or complexity tier, compare each tier's share of files with its share of revenue, assets or flows. If the complex tier is 10% of files but a much larger share of value, "only leaving the complex cases aside" is not a small compromise.

Isn't the happy path already solved?

It should be. A domestic entity with one layer of ownership and good register data can be identified, screened, scored and approved without anyone touching it. Institutions that have automated this path decide simple files in minutes or hours, not days. On Ondorse, a European health insurer ran a periodic review campaign of 5,895 files in October 2026: 73.6% were approved automatically, with no analyst involved. When an analyst does step in, a complete legal-entity file from France or the UK, where registers are well structured, takes 4 to 5 minutes of handling on average (Ondorse production data).

The trouble is what happens at the edge of the happy path. Most onboarding tools assume one entity, one country, one register. When a file breaks that assumption, because of a foreign holding, a missing UBO record or a trust in the chain, the automation stops and the process falls back to its oldest form: an analyst emailing a checklist.

Onboarding becomes all-or-nothing. Either the file is simple enough to automate completely, or it is handled entirely by hand. Your highest-value clients almost always land in the manual group.

What breaks when a complex file falls off the happy path?

Bain's interviews with corporate customers describe the result: "unclear requirements, limited use of existing data, piecemeal requests for documentation" and weeks-long delays before accounts can be used (Bain & Company). The same failure modes repeat across institutions:

• One failed lookup brings back the whole checklist. A single register gap triggers a request for every document, not just the missing one.
• The client is asked for what is public. Certificates of incorporation, director lists and UBO extracts for entities in register-rich countries are requested from the client instead of retrieved.
• Known people are verified again. A director already onboarded for another group entity, or a private client the bank has known for years, goes through identity verification from scratch, even when their file is still valid.
• Requests arrive one at a time. Without a clear view of the whole structure, analysts discover gaps layer by layer, and each discovery becomes another email to the client.
• Data is entered twice. What the client sends is re-typed into the back office by an analyst.

The bill is high. In a 2016 study, Bain put onboarding costs at between $500 and $5,000 per client, with time to cash ranging from 20 to 90 days (Bain & Company). It has not got easier since: in Fenergo's 2025 survey of 600 senior decision-makers, 70% of firms said they had lost clients in the past year because of inefficient onboarding, the highest rate the survey has recorded (Fenergo, 2025).

What does white glove onboarding mean for a complex KYB?

White glove usually means more human attention. Here it means less work for the client: the institution does the work, and the client supplies only the net new, meaning the information no source other than the client can provide.

In practice, the client opens a portal already filled in with the group's entities, directors, ownership chain and official documents, retrieved from registers, ordered from data providers or taken from the institution's own systems. The client checks the structure, corrects what is outdated, and provides the few items only they hold.

Bain's first recommendation to banks makes the same point from the institution's side: "identify what data the organization truly needs to ensure compliance", because requirements and forms "tend to swell over time" (Bain & Company). The net new principle adds a second question for each data point: who is best placed to supply it?

How do you apply the net new principle to a complex file?

Run each element of the file through four tests, in order. The first test it passes determines how the data is collected.

1. Retrieve: Is it in an official register or available from a data provider for this jurisdiction?
2. Reuse: Do we already hold it, still valid, in our CRM, core banking system or another client file?
3. Derive: Can it be extracted from a document we already have, or inferred from another answer?
4. Require: Does this client's risk profile actually call for it?

Only what fails the first three tests and passes the fourth reaches the client. For a typical complex structure, the audit looks like this:

Three design rules keep the audit reliable at scale:

• Map data routes country by country. Register coverage and access vary widely, especially for UBO data. Choose the best route for each data point and jurisdiction, weighing conversion, data quality, regulatory fit and cost.
• Fall back one field at a time. If a route fails for one entity, the client is asked for that one item. The rest of the structure stays pre-filled.
• Let the data trigger the exceptions. Enhanced due diligence steps, country-specific documents and extra reviews appear only when the collected data calls for them.

A worked example

Take an illustrative group: a French holding company (SAS) owns a Luxembourg holding (SARL), which owns 70% of a German operating company (GmbH). There are four legal representatives across the three entities, two of whom are the UBOs. One UBO is a politically exposed person and is already a private banking client of the institution.

Where do complex KYBs concentrate?

Three segments combine high revenue per client with complex structures and little patience for paperwork.

Asset managers onboarding institutional LPs. Institutional investors are often regulated entities whose identity, governance and ownership are well documented in public sources. Asking them to re-send registry documents for every fund they enter slows the fund close. The net new is usually the subscription details, an AML letter and the source of funds. Asset managers also run due diligence on the asset side, covering funds, vehicles and management companies, where the same principle applies.

Private banks onboarding clients through holding companies. The client is a person, but the account belongs to a holding company. The holding's identity, directors and ownership chain can be retrieved and linked to the client's own file before the banker's first meeting. The net new is the source of wealth and the purpose of the relationship, which is exactly where the banker's conversation adds value.

Providers of high-value B2B products such as corporate FX. A corporate treasury team that moves millions around the world has usually been through months of sales discussions before onboarding starts. Group entities, directors and UBOs across jurisdictions can be pre-filled. The net new is the currencies, corridors, expected volumes and signatories, much of which the sales team already knows. UK corporate banks report average onboarding times of more than six weeks (Fenergo, 2025), and in the meantime the trades go to the client's existing bank.

Doesn't asking less weaken due diligence on complex structures?

It focuses it. Complex files often trigger Enhanced Due Diligence (EDD), and the net new principle does not remove any of it. It moves analyst time from re-keying registry data to the judgement calls that matter: unusual ownership chains, source of wealth, gaps between declared and registered structures.

The regulatory requirement is to verify, not to collect everything from the client. And verifying means comparing sources, not trusting any single one. EU AML rules have long required obliged entities not to rely exclusively on beneficial ownership registers for Customer Due Diligence (CDD) (Directive (EU) 2015/849, as amended). The AML Regulation that applies from July 2027 keeps that approach (Regulation (EU) 2024/1624).

The Anti-Money Laundering Authority's (AMLA) final draft technical standards on CDD, submitted to the European Commission in October 2026, set out how to judge whether a source is reliable and independent: its credibility, its official status and independence, how up to date it is, its accuracy, and how easily it can be forged [PLACEHOLDER: link to AMLA final draft RTS on CDD, Article 8]. Data retrieved directly from an official register scores well on most of these. But the same standards treat the central beneficial ownership register as necessary, not sufficient, to verify a UBO. That is why a net new journey cross-checks every source instead of replacing one with another:

• The client confirms every pre-filled data point, so the declaration is still theirs.
• Registry data is compared with the client's declarations and with the documents provided, and every discrepancy becomes an analyst task.
• Every retrieval, confirmation and decision is logged with its source and timestamp.

How do you know complex onboarding is working?

Track your indicators separately by complexity tier. Blended averages let a strong happy path hide a broken complex path.

• Net new ratio: fields provided by the client as a share of all fields, for complex files specifically.
• Documents retrieved versus uploaded: how much of the structure was documented without the client.
• Touches per file: how many times the client is contacted after the first invitation.
• Analyst minutes per complex file: time spent on judgement, not data entry.
• Drop-off rate: the share of complex files abandoned before a decision.
• Time to decision: elapsed days from invitation to approval, the number your sales and treasury teams feel.

How Ondorse handles complex files

Ondorse's portal applies the net new principle beyond the happy path. Entering a registration number triggers register queries that pre-fill the file and download key documents. Where registers stop, UBO extracts are ordered in the background and parsed by AI for an analyst to check, and the client is asked only when every route has failed. Known individuals are reused while their verification is still valid. Requirements specific to a jurisdiction or risk level appear only when the data triggers them.

How much of a group structure can be documented without the client depends on each country's registers. Ondorse's global coverage map shows, country by country, which data and documents Ondorse retrieves automatically, so you can see in advance which parts of your complex files will be pre-filled.

The effect shows up most on complex files. At a payment fintech using Ondorse, analyst review time fell at every complexity level: from 20 to 10 minutes for complex files, from 12 to 5 minutes for medium ones, and to under 5 minutes for simple ones. Its clients' self-onboarding time dropped from between 2.5 hours and 1.5 days to under 10 minutes, and the compliance team it planned to need shrank from 6 to 2.5 FTE. More results are in Ondorse's customer stories.

Automating the happy path should already be done. Your best customers are not on it. If your complex files still run on checklists and email, speak with our team about how Ondorse would handle them.

‍