A data site earns trust by how it handles being wrong. Heriux finds most of its errors by machine, fixes them at the source, and does not quietly pretend they never happened.
A price-integrity check continuously compares current prices against each asset's own recorded history and flags disagreements. A content QA gate blocks broken generated text before publication. An algorithm monitor watches every detector's output distribution and raises an alert when one suddenly changes shape. Most corrections start as one of these alerts, not as a reader complaint.
Errors are fixed at the source — the ingestion pipeline, the generator, the detector — not painted over in one page, so the same mistake cannot resurface elsewhere. Where an error affected published statistics, the statistics are recomputed from the corrected data.
Found something wrong — a price, a statistic, a sentence that misstates the data? Use the contact page. Reports that identify real defects are acted on; the fix lands in the pipeline, and the page corrects itself on the next generation pass.