Signal-to-Noise Ratio

Signal-to-noise ratio borrows a term from engineering to describe the proportion of a dataset that actually informs a decision. It is the number that makes raw volume ambiguous: a source producing a million records a day with a thin ratio costs more to process than it returns, while a smaller, denser source pays for itself. Anyone who has staffed a team to filter a supplier's output has paid the difference in salaries.

Raising the ratio is either the buyer's problem or the provider's. Pushing evaluation upstream - resolving entities, removing duplicates, judging urgency, authenticity, and data quality while a record is still in motion - means what arrives is already the part worth reading, rather than undifferentiated volume with the signal somewhere inside it. The trade to watch is that filtering on the client's behalf destroys information: exposing the evaluation as values on the record lets the consumer set their own threshold instead of inheriting the vendor's.