What breaks first: version conflict?
Two editors, two copies, two decisions: the classic end-of-week merge where nobody knows which gate outcome stands. Repair: a single shared file per register with one editing owner at a time, change history switched on (check your tier’s version-history behaviour rather than assuming it) and filenames that never fork: no “pipeline FINAL v3 (Priya’s edits)”. The definition this discipline serves is on what a qualification layer is.
What breaks second: missing evidence?
Register rows that assert compliance with no linked proof: “ISO held” pointing at nothing, referee names without consent dates, insurance figures from memory. Repair: every claim cell links to its evidence (certificate, email, schedule extract) stored beside the register in a fixed folder per pursuit, with the addendum version each row was checked against. Evidence missing at gate time is a review, not a bid; the same rule as everywhere else on this site.
What breaks third: handoff failure?
The decision made in a meeting, the writer briefed verbally, the unknowns never written down, and three weeks later nobody can reconstruct why this tender was a bid. Repair: the handoff contract in the file (decision, evidence pointers, open questions with owners and dates) filled before writing starts. A reusable decision record is a later template; until it ships, the handoff rows in the register carry the weight.
What does this mean for the bid/no-bid call?
Run the gate from the file, in the file, with history on. The moment decisions happen outside the record (in chat, in corridors, in someone’s head), the spreadsheet system has failed regardless of which platform hosts it. Fix the discipline before buying the tool.