Good solutions
When a flow becomes master data – a line of reasoning
Master data never appears out of nowhere – it is born in a transaction. Here we examine the moment when an ordinary information flow becomes a state the whole organisation depends on.
By Jens Fahlén · Published · 3 min read

All master data begins as movement
It is easy to view master data as something static: a table in a database, a list in a register or a branch in an organisation chart. But in reality, almost all information begins as movement.
A person is hired. A supplier submits an invoice. An employee collects a laptop. A manager approves an order for a consultant.
In the moment, each of these events is simply a flow between two parties or two systems. But at a certain point the information stops being a completed transaction and becomes a persistent state. It becomes master data.
Event or state?
An information flow is an event. A case is handled, an order is delivered or a payment is made. Once the task is done, the flow has served its purpose.
Master data is the entity the event refers to. The supplier who was paid, the employee who was hired or the unit that received the delivery. These records live on long after the transaction is archived.
The problem in many organisations is that flows are built for the immediate need. Only much later does anyone notice that the flow has quietly become the sole source of something ten other systems rely on.
Four signs that a flow has become master data
1. The lifespan outlasts the process. Once the consultancy contract is signed, procurement is done, but the consultant's name, person ID and engagement dates are still needed for login, badges and time reporting.
2. More than two parties depend on the identity. When a third and fourth system start referring to the same object – without knowing where it was created – the information has become a shared point of reference.
3. An error spreads into other domains. If an incorrectly registered manager halts approval flows or gives the wrong people access, the flow is carrying master data.
4. Rules become necessary. Master data needs validation: Does the unit have a valid primary cost center? Is there an active manager code?
The trap: cleaning up afterwards
When the shift goes unnoticed, records start to diverge across registers. Then the idea of an MDM tool often appears, one that will match duplicates and produce so-called golden records.
As we discussed in “The tool that couldn't create agreement”, it is an understandable instinct, but it starts at the wrong end. A central tool can match strings, but it cannot create source ownership for records that were born without rules.
From courier to gatekeeper
Taking control of master data is not about blocking integrations or building huge central repositories. When a flow touches an entity that others rely on, it must act as a gatekeeper.
Before a record takes root in the IT landscape, four fundamentals must be in place:
1. The design is decided. Which building block is this? Which relationships, keys and code lists define the object? Without an agreed conceptual model, the flow only moves technical fields without shared meaning.
2. The source and owner are designated. Only one system can own the truth about the entity, and one designated role in the organisation must have the mandate to approve or reject the record.
3. Quality is checked at creation. The check must happen where the record is born – at registration, in the form or in the flow itself. Letting incomplete information through with a plan to clean it later is the root of information debt.
4. The business rules are enforced. Which conditions must be met before the record counts as valid and may spread further?
When the four parts are in place, information flows stop creating disorder. Each flow instead becomes a way to keep the organisation's shared master data clean and reliable from the very first second.
Tags
- #master data
- #information flow
- #source principle