KYC onboarding is measured almost everywhere by how fast a customer gets through it, which is a reasonable thing to care about and a terrible thing to optimise alone. The output of onboarding is not an accepted customer, it is a file, and a fast process that leaves a thin file has moved the cost rather than removed it. This page is about what onboarding should leave behind.
Decide the rating before you collect
The most common onboarding design collects a fixed set of evidence from everyone and rates the customer afterwards. That gets the order backwards: the rating should determine the evidence set, so standard customers are not over-collected and rated-up ones are not under-evidenced. Doing it in the right order also means the person collecting knows why they are asking for a thing, which is the difference between a request a customer accepts and one they argue with.
Set the review date at onboarding, not later
The review interval comes from the rating, so it is knowable the moment the rating is made. Setting it then, on the record, is the single highest-value habit in this whole subject, because the alternative is a periodic sweep that has to work out what is due, and sweeps are where books quietly go unreviewed. A file created without a next review date is a file that will need somebody to remember it.
Business onboarding is a different clock
KYB onboarding cannot be finished in one sitting when ownership has to be traced, so it needs a state between started and complete, and a rule about what the customer can do meanwhile. Firms that only have open and closed end up either blocking business customers for a week or opening them with an incomplete file and no marker on it. The second is much more common and much worse, because nothing on the record says it was provisional.
Questions people ask about kyc onboarding
How fast should KYC onboarding be?
For standard-risk individuals, minutes. For business customers with layered ownership, days is normal and pretending otherwise leads to incomplete files. The number worth watching is not the average, it is how many files were opened without the evidence the rating required.
Can onboarding be fully automated?
The collection and the checks can be, and plenty of vendors sell that. The rating rationale and the approval on a rated-up file are judgements, and a record of a judgement nobody made is worse than a slower process.
What should the file look like the day after onboarding?
Evidence with sources and dates, a rating with a reason, an approver where the rating needed one, and a review date in the future. If it has those, onboarding did its job.