Your volume
Companies, not individuals — KYB and EDD files. Individual KYC runs on different economics and is not modelled here.
Simple · 1 owner
Standard · UBOs
Enhanced · EDD
Simple: one owner, clean documents. Standard: beneficial owners to trace. Enhanced: cross-border structures, source of funds, documents chased from the customer. They differ threefold in time, so the mix moves the answer more than headcount does.
Salary plus employer contributions, over 1,700 billable hours. Lithuania's rate assumes an open-ended contract in risk group I; Estonia's includes the 33% social tax plus 0.8% unemployment.
What we take off the desk
Set these to what a pilot has actually shown you. They start where the product stands today — not where a sales deck would put them.
Simple
Standard
Enhanced
Set per case type, because only the easy files go through untouched — and simple cases are a third of your count but under a tenth of your hours. An enhanced file never closes itself.
Document reading, party extraction and the first draft of the findings. No independent benchmark exists for this — it is the one number a pilot must measure in week one.
Charged on every case, including the ones that close themselves. The model checks this against what the time is worth rather than assuming you will accept it.
Assumptions you can change
Minutes: a small payments firm runs a standard file in one to two hours — the band this sits in, cross-checked against outsourced KYB prices (€6 routine, €18 high risk, €45 enhanced). At the other end, Oliver Wyman timed corporate onboarding at a large bank at 15.3–25 hours per file. If your files look like the bank's, raise these numbers hard. Data: a registry lookup with beneficial owners starts at $9 and reaches $96 in some countries (Didit's published rates); screening adds $0.20 per company and per person. Utilisation and supervision: 65% and 15% are conventions, not measurements. Together they set the multiplier on analyst time — 1.77×. Fixed cost: MLRO, independent review, tooling and licences. It is shown so you can see it, and deliberately excluded from the saving, because none of it falls when a case takes less time.
Analyst hours returned each year
—
—
True from the week you start. It is capacity, not cash — the hours go back into the queue, the backlog, and the reviews that keep slipping.
Variable cost — the part that moves
| Time saved, valuedhours × loaded rate × supervision multiplier | — |
| Our fee— | — |
| Difference | — |
—
What does not move
| Provider datayou keep your providers; we sit above them | — |
| Fixed compliance costMLRO, independent review, tooling, licences | — |
Anyone who shows you these two lines shrinking is selling you something. They fall only when a whole role or contract goes — see the next panel.
Capacity, and the one way cash follows
| Analyst capacity needed today | — |
| Analyst capacity needed with Evda | — |
| Extra cases the same team absorbsour contribution only, not rounding | — |
| Cash, if a whole post goes— | — |
Hours come back continuously; money comes back in steps. Until a headcount or an outsourcing contract actually ends, the saving is capacity — which is worth having when your queue is the thing losing you customers.
What this rests on
measured Loaded hourly cost from live job ads plus the Emerald Zebra Cyprus survey (Nov 2025, thin sample by its authors' own warning); employer contributions from Sodra (LT — 1.77% applies to open-ended contracts in risk group I) and the Estonian Tax Board (33% social tax + 0.8% unemployment, 2026). Provider data from Didit's published rates. Corporate onboarding timed at 15.3–25 hours per file by Oliver Wyman, via Thomson Reuters' time-and-motion work (c. 2018) — large banks, many hands. Abandonment of just over one application in ten: Fenergo's 2025 survey of 600 executives.
estimated Minutes per case type for a small payments firm, the split of the case mix, utilisation, supervision, the fixed-cost default, and both automation settings. There is no independent measurement of how long a KYB file takes at an EU payments firm, or of how many files an analyst closes in a year — every published figure of that kind we could find comes from a vendor selling automation, usually citing itself.
Deliberately excluded: abandoned applications and lost revenue (we would have to guess your margins), regulatory fines (we will not price a product on fear), and savings on verification providers (we do not replace them). McKinsey's 100-minute review is often quoted for work like this; we do not use it, because it measures periodic reviews of low-risk retail individuals, not onboarding a company.