Guide

How to write a tender response that scores

A tender answer is marked against a published scheme by someone with a stack of submissions and limited time. Almost everything that makes a bid win or lose follows from taking that seriously.

Last checked: 6 October 2026

Find tenders for your company

Paste your website and we score this week’s UK public sector tenders against it, each with a reason. Free, no sign-up.

What is actually happening to your words

An evaluator has your answer, the question, and a marking guide that describes what a high score looks like. They read, they look for the things the guide lists, and they award a mark. Often several evaluators score independently and then moderate.

Two consequences follow, and they explain most losing bids.

  • They score what is written, not what is true. They cannot credit experience you did not describe, even if they know you have it.
  • They are looking for specific things. Anything that makes those harder to find costs marks, however well written it is.

The GOV.UK guidance for SMEs lists the common reasons bids fail: failing to answer the specific question, failing to address the full specification, and insufficient detail. All three are writing failures rather than capability failures.

How quality answers are scored

Under the Procurement Act 2023 (section 23), the buyer must set out its award criteria and an assessment methodology describing how tenders will be assessed against them, and say how important each criterion is: by a percentage weighting, by ranking them, or by describing it some other way. Government guidance on assessing competitive tenders says the methodology includes any scoring matrices the evaluators will use. So the marking scheme is in the documents. Find it before you write a word.

No law fixes the scale. A 0 to 5 scale is a common approach, with a short descriptor for each score. Something like this, though every buyer writes its own:

ScoreWhat it usually means
5Fully meets the requirement, with detailed, relevant evidence and no concerns
4Meets the requirement well, with good evidence and minor gaps at most
3Meets the requirement, but the evidence is general or some points are thin
2Partly meets it: significant gaps, or too little evidence to be confident
1Barely addresses it, or raises serious concerns
0No answer, or an answer that does not address the question

The step that matters is from 2 to 3 to 4, and it is almost always evidence: a named example, a number, a method the evaluator can picture working. Read the buyer's descriptors for 3 and 4 and make sure your answer contains everything they mention.

Weightings turn scores into points. A question worth 10% scored 3 out of 5 usually earns 6 of those 10 points, so a one-point improvement on a 20% question is worth twice the same improvement on a 10% one. Where several people evaluate, the Cabinet Office's bid evaluation guidance describes a moderator leading them to a single consensus score, which should not simply be an average. An answer has to convince a room, not one reader.

Price is scored differently

The tender documents say how price is scored. Many buyers score price relative to the lowest price, though current guidance asks them to be aware of the drawbacks, and a contract can be awarded on price alone. If a price is so low the buyer doubts it, the Act lets it reject the tender, but only after asking you to show you can deliver at that price (section 19).

After the award, every bidder whose tender was assessed receives an assessment summary with its score for each criterion, an explanation of each score, and the same for the winning tender. Reading it properly is the fastest way to move from 2s to 4s.

Use the question as the structure

The single highest-return habit in bid writing. If the question has four bullet points, your answer has four subheadings, in the same order, using the buyer's words.

It feels mechanical. That is the point. The evaluator can find each element without hunting, and cannot accidentally miss one. GOV.UK guidance says the same: adopt the bullet points as subheadings and address each point in the order it appears.

A structure that works for most quality questions

  1. The requirement. One sentence restating what is needed, in the buyer's terms. Shows you understood before you start telling them things.
  2. The method. What you will actually do, in sequence. Tools, governance, who does what. This is the “how”, and it is where most answers go thin.
  3. The people. Named roles, real experience, honest capacity. Not the whole company, the team for this contract.
  4. The controls. What you do when it goes wrong: risks, escalation, quality assurance, reporting, contingency. Buyers are pricing risk as much as service.
  5. The evidence. One concrete example. Client, date, your role, what you did, what changed, with a number.
  6. The outcome. Tie it back to what the buyer said they wanted.

The test for any sentence

Could a competitor write this exact sentence about themselves? If yes, it is not earning a mark. “We have extensive experience delivering complex projects” is available to everybody. “We ran the same migration for Leeds City Council in 2024, moving 1,400 users over three weekends with no unplanned downtime” is not.

Words to delete

DeleteReplace with
Best in class, world class, market leadingThe measurable result that would justify the claim
Highly experienced teamNames, roles, years on this specific type of work
Excellent customer serviceResponse times, satisfaction scores, what you did when a client escalated
Bespoke, tailored solutionThe specific thing you would change for this buyer, and why
We could, we would aim to, we typicallyWe will. Uncertainty reads as a hedge and scores like one

Evidence beats adjectives

The strongest single upgrade to most answers is replacing claims with examples that carry numbers. Not because numbers are impressive, but because they are checkable, and checkable statements are what a marking scheme rewards.

Keep an evidence library: projects with client, date, scope, your role, what changed, and the measure. Reuse that. Do not reuse whole answers, which never quite fit the next question and read like they do not.

Respect the mechanics

  • Word limits are enforced by portals that truncate. Write to the limit, not past it.
  • Answer only what is asked. Material about services not in the specification spends your word count and earns nothing.
  • Define acronyms once. The evaluator may not be from the department that wrote the spec.
  • Make each answer stand alone. Cross-references to another answer often are not followed.
  • Short sentences, real headings, white space. A dense block of text is harder to score and scores worse.

Get someone else to mark it

Before submission, have someone who did not write it score it against the published criteria, pretending to be the evaluator. Not proofread: score. Ask them where each mark is being earned.

If they cannot find the evidence for a criterion in under a minute, neither will the evaluator. That review finds more marks than another day of drafting.

If nobody in-house can do that, a bid review from an outside firm is the cheapest form of help; our directory of bid writing services lists firms that offer reviews, with their published prices.

Find the tenders worth writing for

Tender Radar reads every new UK public sector notice, scores it against what you do and emails you the ones worth reading every weekday morning, with a reason for each. 30 days free, no card needed.

Common questions

Writing answers, in short

Why did my tender answer only score 2 out of 5?

On most scoring scales a 2 means the answer partly met the requirement but left significant gaps or gave too little evidence for the evaluator to be confident. Each buyer publishes its own scale, so read the descriptor for 2 and 3 in the tender documents, then compare your answer with the explanation in your assessment summary, which must explain each score by reference to your tender.

How are tender quality scores weighted?

The tender documents must say how important each award criterion is, by a percentage weighting, a ranking or a description. With percentages, your score is usually converted: 3 out of 5 on a question worth 10% earns 6 of those 10 points. The question weightings tell you where the marks are, so spend your time in proportion.

How should I structure a tender answer?

Follow the question's own structure. If it has bullet points, use them as your subheadings in the same order. State the method, name the people, describe the controls, give one concrete piece of evidence with a measurable result, and connect it to the outcome the buyer asked for.

What do evaluators actually look for?

Evidence against a published marking scheme. They score what is on the page, in the order the question asked for it, and they cannot award marks for things you know but did not write. Anything that makes evidence harder to find costs you marks even when the underlying capability is strong.

Should I reuse answers from previous bids?

Reuse the evidence, never the answer. A library of case studies, methods and measured results saves real time. A previous answer pasted in with the buyer's name changed reads exactly like what it is, and it will not match this question's structure or weightings.

What words should I avoid in a tender?

Anything unverifiable: best in class, world class, highly experienced, excellent customer service, passionate, bespoke. They occupy the word count without earning a mark. Replace each with what you did, for whom, when, and what changed as a result.