The honest answer to “how many documents are lost at handover?”

This question is asked often, and it is where the most numbers get invented.

What we could find: a 2002 NIST estimate for the US capital facilities sector of USD 15.8 billion a year from inadequate data interoperability. Its scope is design, construction and operation, not plant handover, and the figure is old. An estimate attributed to FIATECH: owners spend 2-4% of project cost correcting and entering O&M information, while traditional EPC handover gets less than 0.3%. The "USD 8-16 million on a USD 400 million project" figure is an example multiplication, not an audited average.

What we could not find: what percentage of documents is lost when handing over from EPC to owner, or the average delay in days. It does not exist as a measured rate.

What remains as concrete evidence: the data defects themselves. In a CFIHOS application with Petrobras data, not all mandatory attributes could be mapped: some fields did not exist, some were empty, the plant hierarchy differed. This gives no prevalence rate, but it shows where the problem appears.

What can be done: at the handover gate, count "required ↔ accepted" (tags, attributes, documents), the number left after first-pass validation, open redlines, and the hours the owner spends re-entering. These are a much better measure than a lost-document percentage.

Sources: Gallaher (NIST) 2002; MIMOSA/OpenO&M handbook; CFIHOS application paper (2025); NIST/FIATECH handover guide.