CCI cost disclosure will find every gap in your data

The risk score gets the conference slides. The cost tables will cause more failures. Not because the arithmetic is hard, but because CCI cost disclosure stress-tests something most fund data estates have never been asked for: completeness with provenance.

Transaction costs come with a window register

The standard basis is an annualised average of transaction costs over the previous 36 months. Around that default sits an ordered set of exceptions, and the order matters: a product whose own contractual term is under twelve months uses that term; a genuinely new product uses an estimate basis; a product with 12 to 36 operating months uses its own window with the scaling named in the disclosure rather than silently applied. When more than one condition fires, precedence decides, visibly. The basis for the number is never guessed, and never invisible.

One more trap with a polite face: a provider-calculated 36-month aggregate. Useful input, accepted with a complete evidence set: provider, methodology statement, period, calculation date, coverage. What it is not is a safe harbour. Responsibility for the disclosure stays with the manufacturer. "The number came from the administrator" has never once satisfied a regulator asking why it was wrong.

Entry and exit come from the contract, not the ledger

Disclosed entry and exit figures come from contractual charging terms, presented at prescribed bases: a £10,000 lump sum and £1,000 invested annually, each shown as a percentage and in cash, each computed by its own rule. A flat £250 fee is 2.5% on the lump-sum basis and a very different figure on the regular-premium basis. Tiered structures take the containing tier. Caps disclose when they bind.

What is not a permitted basis: dividing last year's charge receipts by fund assets and calling it the entry cost. The regime wants the terms an investor actually signs up to, as structured facts with a source document, not an average of what happened to be collected.

Zero, not applicable, and unknown are three different answers

A known, approved zero entry charge computes as 0.00% and £0 and completes the table. "Not applicable" is a different answer carrying its own reason. And "we have not confirmed" is neither: it blocks. Most spreadsheet estates hold all three states as the same empty cell. That single modelling failure, the blank that could mean anything, is responsible for more disclosure defects than any formula error we have ever seen.

Anti-dilution: same fund, two right answers

Swing pricing, dilution levies, and market taxes must be classified for what each flow actually is: who pays, who receives, what triggers it, through what mechanism, for what purpose. Under the UK presentation, market taxes and levies the fund pays sit positively inside gross transaction costs, and a benefit received by the fund is never used to floor disclosed costs downward. The EU PRIIPs presentation nets differently. Run the same share class into both regimes, as most cross-border manufacturers must, and you get two different, correct answers from one dataset. If your data model cannot carry that distinction, one of your regulators is eventually reading the wrong number.

And then the confirmations

Distribution adds its own gate: written distributor confirmations of the charges. The subtlety is what fails when they are missing. The cost fact can be complete and approved while the written confirmation is absent; that blocks the distribution output, not the calculation. Two different failures with two different owners. A system that reports both as "costs incomplete" sends your ops team to interrogate the wrong people, twice.

Every one of these requirements is a data field with a source, an effective date, and an approver. None of them is arithmetic.

That is the pattern across the whole cost chapter. The calculations are a few hundred lines of well-tested code. The demand is that every input be a governed fact you can defend three years later. Cost reporting built on spreadsheets fails CCI on provenance long before it fails on maths.