The fastest way to answer a tender is to open the last one. The structure is right, the phrasing is the company's, and most of the clauses are the same. So the bid team copies, and changes the project name, and — three bids later — nobody remembers whether the claim in section 4.2 was ever true for the current product revision.
Two different things in one document
A past response contains the company's voice: how it organises a technical narrative, how it phrases a material statement, what it puts first. That is durable and worth learning.
It also contains claims: this product complies with that requirement. Those were true, at best, for the datasheet revision offered in that bid, under that edition of the standard, with the certificates valid at that time. Datasheets get revised; certificates expire; standards get new editions. A past Comply is a hypothesis about the present, not evidence of it.
Separating them
Tender Answer indexes past responses for structure and phrasing, and drafts new responses in that voice. For evidence it goes to the current library — the pinned datasheet revision, the certificate with its validity date, the current ITP. When a past Comply is a candidate for reuse, it is verified against the current document first. If the current document does not support it, the line is marked Needs engineer and the past claim is listed for the bid manager: section 4.2 claimed X in bid 2025-022; current datasheet does not support it.
Why this matters more over time
Each unverified reuse compounds. A claim copied through five bids has five chances to have become false and five bids' worth of institutional confidence behind it. Breaking the chain once — by making evidence always current — is cheaper than discovering the error in a buyer's clarification round, or after award.
The practical effect
Bid teams keep what they liked about copying the last bid (speed, consistency, voice) and lose what they did not know they were risking. The library's revision and validity tracking does the remembering nobody had time for.