Anyone who tells you they will find everything is telling you something about their sales process rather than their method. Open-source intelligence has real boundaries, and they are not evenly distributed — some questions are answered in an hour, others cannot be answered lawfully at all. Knowing which is which before you commission the work is worth more than any assurance.
The promise is attractive because it removes your uncertainty in one sentence. It should worry you for the same reason.
Three things follow from it. Either the provider does not know where the limits are, which shows in the work. Or they do know and intend to fill gaps with inference presented as fact, which is worse, because you cannot tell from the report which lines are which. Or they intend to cross into closed sources, which transfers the risk to you — the party who commissioned it — and makes the material unusable the moment it has to be shown to anyone.
A report that distinguishes between what was established, what was inferred and what could not be determined is less comfortable to read and considerably more useful to act on.
Some data cannot be obtained lawfully by anyone, at any price. Criminal records, the pre-trial investigation register and conviction data are closed. Bank balances and account movements are closed. Correspondence is closed. Telecom data is closed.
Nobody has a lawful route to those, and anyone offering one is selling a breach with your name on the invoice. Where such a question genuinely decides a transaction, the honest route is a direct request to the counterparty, written into the engagement — which is also why our reports include a set of questions to put to them rather than a claim to have answered everything.
Access also changes over time, in both directions. Ukraine's Unified State Register reopened as open data in January 2026, restoring beneficial ownership. Vehicle records have been closed to third parties since February 2022. Property extracts for legal entities moved from electronic to paper channels in December 2025. A provider whose description of what is available has not changed in three years is not describing reality.
Even inside the law, open sources are uneven. A Ukrainian company's ownership is documentable from a state register. The same question in the Netherlands, Ireland, Cyprus or France is effectively closed to a foreign checker after the Court of Justice ruling in C-37/20, and no amount of effort converts a closed register into an open one.
Some things leave no public trace at all. A verbal agreement between two shareholders. An undisclosed contract. An intention. Where a chain ends in a jurisdiction without a public beneficial-ownership register, what can be produced is a documented inference from surrounding evidence — corporate filings elsewhere, litigation naming the actual decision-maker, procurement records, connected-person patterns — and an honest report labels that as inference rather than as a register entry.
There is also a class of finding that is real but unprovable to the standard you need. Something can be visible enough to change how you negotiate and not solid enough to put in front of a tribunal. Those are different thresholds, and conflating them is how reports become liabilities.
Depth is a function of hours, and hours are what you are buying. A basic check in 4 to 24 hours answers whether a company legally exists, whether it appears on sanctions lists and whether obvious problems are visible. It does not unwind a three-layer offshore chain, because that is not a four-hour task at any price.
This is why turnaround on the deeper levels is agreed at the brief rather than published. A single Ukrainian counterparty is not the same job as a structure running through three jurisdictions, and a fixed promise of "48 hours" across both would be a promise about one of them.
We do not present a guess as a fact. Where something is inferred, the report says so and shows what the inference rests on.
We do not inflate the report. Thirty pages of register printouts is not depth; it is padding that hides where the actual findings are. A finding worth acting on should be visible on the first page.
We do not call a signal a verdict. A company sharing an address with twelve others is a signal worth explaining, not proof of anything. The report distinguishes between what is established, what warrants a question, and what stops a deal.
We do not give legal conclusions. We supply facts and analysis; whether they support a claim, and whether any of it is admissible in a specific forum, is for your counsel and the tribunal.
It means exhausting the lawful sources that bear on your question rather than the ones that are quick. It means marking each finding with a confidence level, so you can tell a register entry from a media report from a pattern. It means writing "not established" where nothing was established, instead of leaving a gap that reads as clearance. And it means turning the unanswerable parts into specific questions for the counterparty, so the gap becomes something you can act on rather than something you carry unknowingly.
A report that overpromises costs you twice. Once when you act on a finding that was inference dressed as fact. And again when the material is examined by a bank compliance officer, an auditor or opposing counsel, and the parts that do not hold up cast doubt on the parts that do.
A report with stated boundaries can be used openly. You know which lines are solid, which need a question put to the counterparty, and which are simply not knowable from open sources. That is a smaller claim, and it is the one that survives being read by someone who is looking for holes.
Related reading: where the lawful boundary runs in practice, in is counterparty due diligence legal, and what an evidentiary standard requires, in due diligence evidence standards.
A scoping conversation costs nothing: we say which depth your case actually needs, where a cheaper level answers the same question, and where open sources cannot answer it at all.