Here’s a sentence that appears, in some form, in almost every losing tender we review:
“We are committed to delivering a best-practice, industry-leading service.”
It sounds fine. It’s also worth zero marks. There’s nothing in it an evaluator can score, because there’s nothing in it that’s verifiable. And most bidders have no idea how much of their submission is made up of exactly this kind of sentence – because to the person who wrote it, it reads as confidence, not as a gap.
This is the evidence problem in tendering: the absence of proof is invisible to the author. You know your business delivers; the claim feels true when you write it. The evaluator, who has never met you and can only mark what’s on the page, sees an unsupported assertion. The scoring guidance most public buyers use makes this explicit – top marks go to responses that are “fully supported,” and a response can only be fully supported by evidence.
How AI can find the gaps a tired bid team can’t
This turns out to be a problem AI is genuinely well suited to, because missing evidence leaves detectable fingerprints in the text. BidReview hunts for them in two dedicated stages of its evaluation pipeline.
The failure signatures stage scans for the language patterns that reliably signal an evidence vacuum:
- Empty buzzwords — “best practice,” “world-class,” “industry-leading” — used without a named proof point attached. Each one is a flag saying a claim was made here and nothing backs it up.
- Hedge language — “we would endeavour to,” “where possible,” “subject to” — which signals a commitment the bidder isn’t confident enough to evidence.
- Future-tense promises standing in for past-tense proof. “We will implement robust quality processes” is a hope; “we reduced defect rates by 34% on a comparable contract” is evidence.
The quality guardrails stage then applies the harder test: the me-too litmus. For every claim in your response, ask – could a competitor write this exact sentence? If yes, it differentiates nothing and proves nothing. Claims that survive the litmus test are the ones tied to your specific contracts, your named clients, your measured outcomes.
Each flag is mapped to the criterion it sits under, which matters enormously: an evidence gap in a criterion worth 30% of the marks is an emergency; the same gap in a 5% criterion is a footnote.
Why this matters more than ever
Almost everyone now drafts with AI, and AI-assisted drafts are fluent, structured, and compliant — which makes the evidence gaps even harder to spot by eye. The proposal reads well. But fluency without proof hits a hard ceiling: in our calibration model, a complete methodology with no named proof points predicts 65–70% and cannot exceed 70%. Scores above 79% require multiple concrete proof points. A winning bid typically scores around 90%.
In other words: the entire distance between “we submitted a decent bid” and “we won” is paved with evidence. The bidders who win aren’t necessarily more capable — they’re the ones who replaced every “world-class” with a number, a name, and a date.
A practical exercise for your next bid
Take your strongest answer and highlight every sentence an evaluator could actually verify. For most teams, the result is sobering – and it’s exactly the picture a structured review makes systematic: every unsupported claim, every hedge, every me-too sentence, located, scored for impact, and queued for fixing in priority order.
BidReview’s failure-signature and quality-guardrail analysis shows you precisely where your submission is asserting instead of proving — before the evaluator does. If you’d like that picture for your next bid, get in touch.
