What evaluators actually score — the 6 things bid writers miss

A common bid writer's mistake: assume the evaluator reads like a human. They don't. They read against a scoring matrix, often line by line, and they're looking for specific evidence to tick specific boxes.

This post is the result of (a) reading published evaluation reports for failed bids and (b) talking to current and former evaluators in central government and the NHS. The six patterns below come up over and over.

1. The matrix is the whole game

Every public sector tender publishes its evaluation criteria — usually in the ITT (Invitation to Tender) or the Specification. Read it first, before you write a single word.

You're looking for:

• The scoring scale (often 0–5, sometimes 0–10, occasionally pass/fail)
• The weightings per question (technical 60 / commercial 40 is common but never assume)
• The scoring descriptors ("5 = comprehensive, clear, fully evidenced…")

Your bid is being graded against those descriptors literally. If a "5" requires "specific examples with measurable outcomes" and your answer has no specific examples, you cannot score 5. The evaluator may agree your answer is good — they still can't legally award you a 5 against the published descriptor.

2. The opening line of every question matters

Evaluators read a lot of bids. They're scanning for compliance signals in the first 2–3 lines. Open with:

• A direct restatement of what you're committing to
• The specific evidence you're about to present

Bad opener: "At [Company], we have decades of experience delivering excellent services to the public sector…"

Good opener: "We will deliver [specific service]. Our evidence: three NHS Trust contracts of comparable size delivered to spec in 2022–24, detailed below with measurable outcomes."

3. Specific numbers beat eloquent prose every time

A measurable claim with a date and a source beats a paragraph of adjectives. Compare:

"We deliver outstanding customer service across all our public sector accounts."

vs.

"Our 2024 NHS service desk averaged a 96% first-time-fix rate (1,432 tickets, Q1–Q4) against a contractual target of 90%. Source: NHS England quarterly performance report, attached as Appendix B."

The second one scores. The first one doesn't, regardless of how true it is.

4. Hedging language quietly kills you

Phrases evaluators flag as red:

• "We aim to…"
• "We endeavour to…"
• "Where possible, we will…"
• "Best efforts will be made…"

These read to an evaluator as "we're not committing." The contract is a commitment instrument. Hedging in the bid implies hedging in the delivery. Replace every hedge with a firm verb: "We will." "We deliver." "Our team has."

5. The social value question is half the bid for many contracts

Since PPN 06/20 came in (Jan 2021), social value carries a minimum 10% weighting on central government contracts, often more. Many bidders treat it as a tick-box. The winners don't.

Strong social value answers:

• Quote specific KPIs from the buyer's Social Value Model (MAC themes)
• Promise measurable outcomes tied to those KPIs (e.g. "We will create 4 apprenticeships in [region] within 12 months of contract start")
• Include a methodology for measuring and reporting the outcomes
• Reference real, comparable past delivery — not generic CSR

If you're winging this section, you're leaving 10–20% of the score on the table.

6. The compliance matrix is what gets you to the evaluation

Before evaluators score, a compliance officer checks that the bid meets every mandatory requirement: insurances, accreditations, format, page limits, etc. Bids that fail compliance never reach scoring.

The most common compliance failures:

• Page count exceeded — counted strictly, including appendices unless explicitly excluded
• Word count exceeded — same rule
• Missing accreditation — Cyber Essentials Plus is the new one to watch
• Wrong format — submitted as PDF when DOCX was required, or vice versa
• Insurance evidence stale — public liability and PI must be in date at submission

A 30-minute pre-submit checklist catches all of these. Skipping it is how strong bids lose to weaker ones.

A working pre-submit checklist

• Every question opens with a direct commitment + a pointer to evidence
• Every claim has a specific number, date, and source
• No hedging verbs left ("aim to", "endeavour to", "where possible")
• Social value answer maps to a specific MAC theme + measurable KPI
• Word/page counts checked per question (including appendices)
• All required accreditations listed and in-date
• Insurance evidence current
• Format matches ITT spec (file type, page size, font, naming convention)
• Each answer passes the "score 5 against the published descriptor" test

The 80/20

Three things will lift your win rate more than anything else:

1. Read the evaluation matrix before writing. Most bidders skip this.
2. Quote specific numbers, dates and sources for every commercial claim.
3. Take the social value question seriously — it's worth 10–20% of the score.

Everything else is detail.