Data verification confirms that data was accurately transcribed or transferred. Validation checks whether data is correct at entry. Verification checks whether it arrived intact. A clerk types an address from a paper form into a database. Validation ensures the postal code matches the city. Verification ensures the address was typed correctly in the first place. The two are related but distinct. One checks the data against rules. The other checks the data against its source.

Verification methods vary by context. Double entry has two people enter the same data independently and compares the results. Any discrepancy triggers a review. This is common in clinical trials and financial systems where errors are costly. Hash verification compares a checksum before and after transfer to confirm the data was not corrupted in transit. Record counts verify that a migration moved the expected number of rows. Sampling verifies a subset of records against the source. Each method catches different errors. Double entry catches transcription mistakes. Hashing catches transmission errors. Record counts catch incomplete transfers. Sampling catches systematic problems. Verification is not free. It takes time and effort. The question is whether the cost of an error justifies the cost of verification. For a marketing list, maybe not. For a patient's medication dosage, absolutely.

Verification methods

• Double entry — independent entry by two people
• Hash verification — checksum comparison before and after transfer
• Record counts — confirm expected number of rows
• Sampling — spot-check records against the source
• Reconciliation — compare totals across systems

Verification is a second pair of eyes. It catches what validation misses.