Editorial
An Algorithm Should Not Replace a Decision: What Ukrainian Music Services Can Learn from the MYFOP UA Case
When a service handles money, an automated recommendation should shorten the search without obscuring the logic behind the calculation. The MYFOP UA case shows how to apply this to royalties, advertising expenses, and streaming payouts.
VECTOR describes a case involving the founder of MYFOP UA, an accounting service for individual entrepreneurs: the service was correcting an error related to refunds. The system automatically finds potentially related payments, but does not treat its conclusion as final—the user remains responsible for confirming it.
For products that work with financial data, this distinction is important. An algorithm can quickly narrow down the options, but it should not create the impression that its recommendation is already a fact.
Show not only the amount, but also how it was calculated
It is not enough for a user to see the final result. They need to understand which transactions or data informed the calculation, why the system linked certain operations, and which alternatives it considered.
This is especially important for music products. In a royalty report, the final amount may depend on specific releases, territories, platforms, or payout periods. In advertising analytics, it may depend on which expenses the system attributed to a campaign. If the logic is hidden, the team cannot quickly distinguish a data error from a correct but unintuitive calculation.
A recommendation and a confirmed decision are different states
The interface should clearly indicate what the system did and what the user confirmed. An automatically identified payment is a suggestion for review. The final classification or change to the total should occur only after a specific action by a person.
This approach does not eliminate automation. On the contrary, it gives automation a strong role: the system removes the routine search, while the user checks the context and is responsible for the final choice. At the same time, the team can see at which stage the discrepancy arose and what exactly needs to be corrected.
What this means for music teams
Labels, managers, and services should expect accounting tools to provide not only the final figure but also an explanation of it. In distribution and streaming payout reports, useful elements include a list of data sources, visible assumptions, alternative classification options, and a history of changes made after confirmation.
The practical principle is simple: automate the search, but do not automate trust. Any system that affects an artist’s money should show the grounds for its recommendation, allow it to be checked, and explicitly record the moment when a suggestion becomes a decision.