Ask any bid writer what makes a strong answer, and they'll say the same word: evidence. Ask them to define it, and the answers get vaguer. In practice, most tender responses only ever reach the first of three distinct levels of evidence - which is exactly why they plateau in the middle of the scoring band, no matter how much effort goes into the writing.
Here's the framework, from weakest to strongest.
Level one: the assertion
This is the evidence level almost every bid defaults to. "We have a robust quality assurance process." "Our staff are highly trained." "We take a person-centred approach." These are claims about the organisation, made by the organisation, with nothing behind them but the confidence of the sentence itself. An evaluator reading a level-one claim has no way to verify it, no way to distinguish it from an identical claim in a competitor's bid, and - critically - no reason to award it marks above the middle of the band. Level one is where most tenders stop, not because bidders don't have anything better, but because level one is what naturally comes out when you write the way you'd speak.
Level two: the named, measured process
Level two replaces the assertion with a governed process: a named role, doing a specific thing, on a defined frequency, measured against a target. "The Quality Manager audits 10% of case files monthly, logging any exception for closure within ten working days." This is dramatically stronger, because it's falsifiable - it describes something specific enough that an evaluator can imagine checking whether it's true. Most competent, well-organised bids reach level two on their headline claims. It's a real and meaningful improvement over level one, and it's still not where the marks at the very top of the band are won.
Level three: the outcome, with a name and a date
The top level of evidence isn't the process - it's proof that the process produced a result, in a specific, real, dateable instance, that improved something measurable for someone. Not "we monitor incidents closely," but a specific incident, on a specific service, where the monitoring process caught something, what was done about it, what changed afterwards, and what was learned that's now been built into how the service runs. This is the level that closes an evaluator's discretion - the level at which there's genuinely nothing left to doubt, because you're not describing a capability, you're demonstrating one that has already happened.
Why almost nobody reaches level three
Level three evidence requires something level one and two don't: a live, maintained library of real, anonymised, dated case examples, indexed by theme, ready to be deployed against whatever the next tender asks for. Most organisations don't have this, because building it isn't part of delivering the service - it's a distinct, deliberate discipline, done between bids, not during them. It's also, not coincidentally, the single highest-leverage thing an organisation can build if it intends to tender regularly. The bid that can reach into a real case example, on demand, matched to the cohort and the question being asked, is playing an entirely different game to the bid still writing in assertions.
A practical starting point
You don't need hundreds of examples. You need a handful - one or two per major theme your tenders tend to cover - written up properly, once, with the context, the challenge, the action, the measured result and the lesson learned, ready to be pulled and adapted whenever the right question appears. Most organisations already have these stories. They're just sitting in institutional memory instead of on the page, at exactly the moment they'd earn the marks.