How UK tender responses are actually scored.
The rubric is published in every ITT, yet most bidders write as if a sympathetic reader will fill in the gaps. There is no sympathetic reader. There is a panel with a scoring guide, twenty submissions, and instructions to mark only what is on the page.
The machinery: criteria, weightings, and a scoring scale
Almost every UK public-sector evaluation works the same way. The buyer publishes evaluation criteria — typically a quality/technical component and a price component, weighted against each other (60/40 and 70/30 quality-to-price splits are common for service contracts). The quality component is broken into questions, each with its own weighting. Each question is marked on a fixed scale, usually 0–5 or 0–4, with published definitions for each mark.
A typical 0–5 scale looks like this:
| Score | Standard definition (paraphrased) |
|---|---|
| 5 | Excellent — fully addresses the question with strong, specific evidence; adds relevant value beyond the requirement; no reservations. |
| 4 | Good — fully addresses the question with credible evidence; minor gaps only. |
| 3 | Acceptable — addresses the requirement; evidence present but limited, generic, or partly unsubstantiated. |
| 2 | Minor reservations — addresses parts of the question; material gaps or weak evidence. |
| 1 | Serious reservations — largely fails to address the question. |
| 0 | Unacceptable — no meaningful response, or fails a mandatory requirement. |
Read the definition for 3 again. "Addresses the requirement, evidence limited or generic" is a description of most professionally written bid responses. That is why 3 is the default score for a competent answer — and why bids full of competent answers lose.
Why "most advantageous tender" changed the game less than you'd think
Since the Procurement Act 2023 came into force in February 2025, buyers award to the "most advantageous tender" (MAT) rather than the old "most economically advantageous tender" (MEAT). The vocabulary changed; the mechanics above did not. Questions, weightings, scored scales and moderation panels all survived intact — and legacy procurements and many live frameworks still run under the old Public Contracts Regulations 2015 rules anyway.
What actually happens in the evaluation
- Compliance gate first. Before anything is scored, submissions are checked for mandatory requirements — missing certificates, unsigned declarations, exceeded page limits, unanswered mandatory questions. Failures here can exclude a bid without a single word being evaluated.
- Individual scoring against the rubric. Each evaluator scores independently, question by question, writing a justification for each mark. Evaluators are explicitly instructed to score only what is written — not reputation, not incumbency knowledge, not what they privately know you can do. If the evidence is not on the page, it does not exist.
- Moderation. The panel meets to agree consensus scores. Where evaluators disagree, the discussion anchors on the score definitions — and the consensus tends to settle on the mark that is easiest to defend in writing. An answer that one reader loved and another found vague usually moderates down, because "limited evidence" is easier to justify than enthusiasm.
- Weighting and arithmetic. Consensus scores are multiplied by question weightings, quality is combined with the price score, and the ranking falls out of a spreadsheet. By this stage the outcome was decided paragraphs ago.
The habits that cost marks, in the panel's own terms
- Claims without evidence. "We are experienced in contracts of this nature" is a 3. A named contract with values, dates, KPIs achieved and a referee is a 4 or 5. The scale definitions literally turn on the word "evidence".
- Answering the theme, not the question. Questions often contain several explicit sub-parts ("describe your approach to mobilisation, including TUPE, timescales and risk"). Miss one sub-part and the definition of 4–5 ("fully addresses") is out of reach regardless of quality elsewhere.
- Making the evaluator work. Weightings tell you where the marks are; structure your answer so each sub-criterion is findable in seconds. Buried evidence scores like absent evidence, because a panel marking twenty bids will not excavate yours.
- Coasting on the "soft" questions. Social value (mandated at a minimum 10% weighting in central government since PPN 06/20), sustainability, safeguarding, contract management. These are scored on the same scale as your methodology, and they are where otherwise-strong bids quietly bleed the margin.
- Boilerplate the panel has already read twice today. Evaluators see the same recycled paragraphs across bids — and increasingly, the same unedited AI-generated ones. Generic text cannot earn the "specific, relevant evidence" the upper scores require.
Score your own draft before the panel does
Everything above is public and predictable, which means you can apply it to your own draft before submission: take the published criteria, mark each answer against the 0–5 definitions as coldly as a stranger would, and weight the result. Done honestly, this predicts most of the panel's verdict. The difficulty is the "honestly" — the author of a bid is the worst-placed person alive to notice which claims are unevidenced, because they know what they meant. That blind spot is precisely what buyer feedback will later describe, uselessly, as "insufficient detail" — see our guide to decoding tender feedback, and if the contest is already over, what a debrief must legally tell you.
A mock evaluation, before the real one.
Fernbridge scores your draft the way the panel will — against the tender's own criteria, weightings and score definitions — and returns the full scorecard with the three weaknesses most likely to cost you the contract. £99 fixed, 48 hours, openly AI-performed, refunded if it tells you nothing.
Request a review — £99