There is a persistent myth in the data business that more is always better. Bigger databases, more records, more fields. In practice, the organizations that get real value from data tend to care about something else first: knowing exactly where each piece of information came from.
What provenance means
Data provenance is the documented history of a piece of data. Where was it obtained? When? By what method? Has it been transformed, merged or corrected since? Who is responsible for it? Under what terms may it be used?
These questions sound bureaucratic until something goes wrong. A client acts on a record that turns out to be outdated. A model is trained on data that was never licensed for that purpose. A regulator asks how a dataset was assembled and nobody can answer. In each case, the cost of missing provenance far exceeds the cost of having recorded it.
Volume without provenance is noise
A large dataset with unknown origins is difficult to trust and therefore difficult to use. Every decision built on it carries hidden risk. A smaller dataset with clear provenance, by contrast, can be relied on, audited, corrected and combined with other sources confidently.
This is particularly true for business relationship data. Information about companies, sectors and connections changes constantly. Firms open and close, merge, change leadership, move into new markets. Without a record of when information was last verified, there is no way to know how much to trust it.
Freshness is part of provenance
Every dataset begins to decay the moment it is created. A useful practice is to attach a verification date to every record and to define, for each type of information, how long it can be trusted before it must be checked again. Contact details and leadership roles decay quickly. Founding dates and registration numbers change rarely.
Being honest about freshness is a form of respect for the people who use the data. Presenting old information as current is one of the quieter ways data providers lose trust.
Methods should be explainable
Users of data should be able to understand, at least in general terms, how it was produced. Was it gathered from public registries, from direct relationships, from surveys, from automated extraction? Was it reviewed by people? What are the known gaps and biases?
This does not require revealing every proprietary technique. It does require enough transparency that a careful user can judge whether the data is fit for their purpose.
Usage rights are part of the record
Provenance also includes the terms under which data may be used. Data obtained for one purpose may not be appropriate for another, either legally, under laws like Brazil's LGPD, or ethically. Recording these terms alongside the data, and enforcing them in licensing agreements, protects everyone involved.
It also prevents a common problem: the same information being sold or shared repeatedly to parties who each believe they have exclusive access. Clear records of who has licensed what, and on what terms, are essential to honest dealing.
Building a provenance culture
Provenance is ultimately a cultural commitment as much as a technical one. It requires everyone who touches data to record what they did and why. Simple tools are enough to start: a source field, a date field, a method field and a usage terms field on every record, plus the discipline to fill them in.
At IGMEP Data, provenance is central to how we think about our work. Our value comes from relationships built over many years, and the information connected to those relationships is only valuable if we can explain where it came from and stand behind it. Volume is easy to claim. Provenance has to be earned.
Existe o mito de que mais dados é sempre melhor. Na prática, quem extrai valor real dos dados se preocupa primeiro com outra coisa: saber de onde cada informação veio.
O que é proveniência
É o histórico documentado do dado: origem, data, método, transformações, responsável e termos de uso. Parece burocracia até algo dar errado.
Volume sem proveniência é ruído
Uma base menor com origem clara pode ser auditada e combinada com confiança. Frescor também faz parte da proveniência: cada registro deve ter data de verificação.
Direitos de uso
Os termos de uso fazem parte do registro, protegem todos os envolvidos e evitam que a mesma informação seja vendida repetidamente a quem acredita ter exclusividade. Volume é fácil de alegar. Proveniência precisa ser conquistada.