How to write a tender bid: question by question
A tender response is not one document. It is a set of separately scored answers, each with its own criterion, weighting and evidence requirement. Here is what each type is scored on.
CEO & Co-Founder
On this page
How to write a tender bid comes down to one structural fact: a bid is answered question by question, not written as a single document. Each question has its own award criterion, its own weighting and its own evidence requirement.
Each answer is scored separately against a published methodology. Anyone who has written one knows the moment the document arrives and the instinct is to read it front to back like a brief. It is not a brief. It is a scoring sheet with prose wrapped around it, and the questions in it are not variations on a theme - a capability question and a methodology question are assessed by different people looking for different things, and an answer that would score well against one will score badly against the other. That is how a competent bid loses marks: by answering the question that was almost asked.
How is a tender response actually structured?
Knowing how to write a tender response starts with knowing which part of it you are in. A tender response has three layers, and they are usually written by different people under different pressures. There is the selection and compliance layer: the conditions of participation, the declarations, the certificates and the portal fields. There is the answer layer: the scored questions, which is where the marks are. And there is the pricing layer, which is scored separately and almost always contradicts the answer layer unless somebody deliberately reconciles the two.
The first layer carries no marks, which is exactly why it gets left to the last afternoon, and it eliminates more bids than the answers do. It works on two different footings and the difference matters. Conditions of participation are the gate: an authority may set them only where they are a proportionate means of ensuring that suppliers have the legal and financial capacity or the technical ability to perform the contract (source: legislation.gov.uk s.22(1), checked September 2026), and it must disregard any tender from a supplier that does not satisfy them (source: legislation.gov.uk s.19(3)(a), checked September 2026). That one is mandatory and there is no discretion in it. Everything else in the layer is discretionary: an authority may disregard any tender that breaches a procedural requirement set out in the tender notice or associated tender documents (source: legislation.gov.uk s.19(3)(d), checked September 2026), and the Act is explicit that a supplier breaching a procedural requirement counts for this purpose, not only a defect in the paperwork (source: legislation.gov.uk s.19(6), checked September 2026). A missed word limit or an unsigned declaration is not a presentational problem. It is a stated ground for putting the whole submission aside without reading it.
The answer layer is where this article lives. Each scored question maps to an award criterion, and the criteria are not free text: they must relate to the subject-matter of the contract and be sufficiently clear, measurable and specific (source: legislation.gov.uk s.23(2), checked September 2026). That obligation is the bid writer's best friend, because a criterion that has to be measurable is one the authority has had to describe: in setting the criteria it must describe how tenders are to be assessed against them (source: legislation.gov.uk s.23(3)(a), checked September 2026). The thing being measured exists in the documents you already have.
What is each question type scored on?
Tender questions repeat across contracts, sectors and buyers, and across formats: the same types turn up in an RFP response under different headings. There are seven that recur, and each has a different thing the evaluator is actually looking for, a different evidence expectation, and a different way of going wrong. The table below is the one worth keeping open while you write.
| Question type | What the evaluator is scoring | Evidence to attach | Why this answer type loses marks |
|---|---|---|---|
| Capability | Whether you have done this specific thing at this scale, not whether you could | Named contracts of comparable size and sector, as many as the question asks for, with dates and values | Capability described in the abstract, with no named contract behind it |
| Methodology | Whether the method is specific to this contract and survives its constraints | A named process, the roles that run it, and the points where it is checked | The standard process with the buyer's name pasted into it |
| Social value | Measurable commitments against the buyer's chosen themes and outcomes | Commitments with numbers, timescales and a named accountable owner | Aspirations rather than commitments, with no way to measure delivery |
| Pricing narrative | Whether the commercial model is consistent with the solution, since price itself is usually formula-scored | Assumptions, exclusions, and the link to the resourcing in the method statement | A price that quietly contradicts the resourcing the methodology promised |
| Risk | Whether you have identified the risks that apply to this contract specifically | A register with owners, mitigations and residual scores | A register whose top risks belong to the sector rather than to this contract |
| Mobilisation | Whether the timetable is credible and the people in it genuinely exist | A dated plan to go-live with named roles and the dependencies between them | A plan that quietly assumes the buyer will hit every one of its own dates |
| References | Whether the referee will confirm what you claimed | Contactable referees who know the contract and were briefed before submission | Referees who were never told, or who cannot speak to the claim made |
The fourth column is the useful one. Read down it and the pattern is the same failure repeated in seven costumes: an answer that is true about the organisation but not anchored to anything the evaluator can check. Evaluators are not scoring whether you are good. They are scoring whether the answer in front of them demonstrates it, using only what is in front of them. The pricing row is the one with a statutory safety net attached: an authority may disregard a tender whose price it considers abnormally low, but before doing so it must notify the supplier and give it a reasonable opportunity to demonstrate that it can perform the contract at that price (source: legislation.gov.uk s.19(3)(c) and s.19(4), checked September 2026). If the supplier demonstrates it to the authority's satisfaction, the tender may not be disregarded on that ground (source: legislation.gov.uk s.19(5), checked September 2026). That opportunity is only worth having if the bid already explains how the price was built.
Bid writing examples: the same answer, written weak and written strong
Most bid writing examples show a finished answer and leave you to work out why it scored. The pair below is the same answer twice. Take a capability question: describe your experience of delivering a comparable service. The weak version is the one almost everyone writes first, and it reads like this. “We have extensive experience delivering services of this type across the public sector and have a proven track record of successful delivery.” Every clause of that is unverifiable, so an evaluator has nothing to award against and the answer sits at whatever the descriptor calls partial. The strong version makes the same claim checkable. “We deliver the same service for two comparable authorities, one over three years to 2025 and one running now, both within ten per cent of this contract by value and both named in our references. The transferable elements are the mobilisation model and the reporting cadence. The difference here is the multi-site footprint, which we cover with the regional supervisor structure set out in our method statement.” Same organisation, same track record, same word count. The second one names things an evaluator can check, and it volunteers the gap rather than leaving it to be found.
In what order do you write a tender response?
Most guidance on how to write a tender walks the document from front to back, because that is the order it arrives in. It is the most expensive order available. The compliance layer is quick and mechanical, the heavy questions need source material that takes days to assemble, and the pricing cannot be reconciled against a method statement that has not been written yet. Working in document order means finding all three out too late. The sequence below is the one teams who do this every week converge on, and it is driven by the mark scheme rather than by the pagination.
- Read the award criteria and the assessment methodology first, before a single question. The authority has had to describe how tenders are to be assessed against the criteria (source: legislation.gov.uk s.23(3)(a), checked September 2026), so the scoring logic is already written down somewhere in the pack.
- Build an answer plan from the mark scheme rather than from the question list. One row per scored question, carrying its weighting, its word limit, its evidence requirement and the name of the person who owns the source material.
- Rank the questions by weighting against the bands they could realistically move. A criterion carrying 30 per cent and sitting one band below the top is worth more attention than one carrying 5 per cent and already at the top, and that ranking is what stops the week being spent evenly.
- Raise clarifications on day one rather than in the last week. A criterion that is not clear, measurable and specific is a question to ask (source: legislation.gov.uk s.23(2), checked September 2026), the answer is normally published to every bidder, and the clarification window closes long before the deadline does.
- Pull the source material and date-check it before drafting anything. Certifications, insurance, accounts, reference contracts and case studies all expire, and on most bids the checking takes longer than the writing.
- Draft each answer against its sub-parts, in the order they were asked. Then reconcile the pricing narrative against the resourcing the method statement has just promised, because that contradiction only becomes visible once both exist.
- Run the compliance layer last, and never on the last afternoon. Word limits, declarations, signatures, mandatory attachments and portal fields carry no marks and can still have the whole submission disregarded (source: legislation.gov.uk s.19(3)(d), checked September 2026).
Two of those run against instinct. Clarifications feel like something you raise once you understand the questions, and they are worth raising before you do. Compliance feels like a final tidy, and it needs a day of its own, checked by someone who did not write the answers.
How do you write to the scoring criteria?
Read the methodology before you read the questions. Teams that spend their year writing tenders do this first and everyone else does it last. Under the Procurement Act 2023, in force since 24 February 2025 under the Commencement No. 3 Regulations, SI 2024/716 (source: legislation.gov.uk, checked September 2026), a contracting authority must describe how tenders will be assessed and must specify whether failure to meet a criterion would disqualify it (source: legislation.gov.uk s.23(3)(a), checked September 2026), and, where there is more than one criterion, it must indicate their relative importance, either as percentage weightings, as a ranking, or described in another way (source: legislation.gov.uk s.23(3)(b), checked September 2026). Those two facts change what you write. A criterion that disqualifies on failure is not a place to be interesting. A criterion carrying 5 per cent does not deserve the afternoon that a criterion carrying 30 per cent needs, and bid teams routinely give both the same afternoon because both are one box on the portal.
And a criterion that is not clear, measurable and specific is a question to ask, not a gap to write around. The clarification window exists for exactly this, the answer is normally published to every bidder, and the obligation in s.23(2) is what entitles you to press for one. A question that forces an ambiguous criterion into writing is worth more than an afternoon of careful drafting against a guess.
The criteria can change after you have started
In a competitive flexible procedure the authority may refine an award criterion part-way through, provided the tender documents allowed for the refinement and it has not yet invited tenders for final assessment, and it must then modify and republish the documents the refinement affects (source: legislation.gov.uk s.24(1) and s.24(4), checked September 2026). One limit runs in the suppliers' favour: a refinement may not produce criteria under which a supplier already eliminated at an earlier round would have progressed (source: legislation.gov.uk s.24(3), checked September 2026). The timing is the part worth understanding, because it cuts both ways. A criterion cannot be refined once the authority has invited the tenders it will actually score, so this will never land in the middle of drafting your final submission. What it can do is move the ground between rounds of a competitive flexible procedure, and it arrives as a document reissue in a portal notification that looks purely administrative. Diff every reissued document against the version you were writing to, on the day it arrives.
Social value is a scored section, not a closing paragraph
For central government departments, executive agencies and non-departmental public bodies, in-scope organisations must apply a minimum 10 per cent weighting, or an equivalent measurement, of the total score for social value, and this applies to procurements commenced under the Procurement Act 2023 on or after 1 October 2025 (source: gov.uk PPN 002, checked September 2026). Where an absolute methodology such as price per quality point is used, that 10 per cent applies to the quality score rather than to the total (source: gov.uk PPN 002, checked September 2026). The floor reaches above-threshold procurement only, and it does not reach private utilities contracts or Ministry of Defence defence and security contracts (source: gov.uk PPN 002, checked September 2026). Below threshold, inside those exclusions, or outside central government altogether, the weighting varies, so read the methodology rather than assuming a floor applies. Where it applies, social value is worth more than most single technical questions, and it is still commonly written last, by whoever is free, from a template. That is a scoring decision disguised as a resourcing one. That floor is already scheduled to move. Under PPN 026, published 5 August 2026, in-scope organisations should apply a minimum 10 per cent weighting to contracts valued at GBP 1 million or above but under GBP 5 million, and a minimum 20 per cent weighting to contracts valued at GBP 5 million or above, for procurements commenced on or after 1 January 2027 (source: gov.uk PPN 026, checked September 2026). The same note narrows the scope to covered procurements worth GBP 1 million in total contract value including VAT or above (source: gov.uk PPN 026, checked September 2026). If you are writing for a contract that goes out after the turn of the year and it is worth more than GBP 5 million, social value is about to carry as much weight as most whole technical sections.
The descriptors are where the marks hide
Most mark schemes publish a descriptor for each score, and the descriptors are where the marks actually hide. The middle of the scale is normally reserved for an answer that meets the requirement in full, and the top for one that meets it and evidences something beyond it. That has a practical consequence for how the week gets spent. Moving a heavily weighted question from meeting the requirement to exceeding it is worth more than perfecting a lightly weighted one, and the distance between those two bands is almost always measured in evidence rather than in prose. Work out which questions can realistically move a band before you allocate anyone's time.
Answer the question that was asked, in its own order
Evaluators score against a mark scheme, usually one that mirrors the wording of the question and its sub-parts. Where a question has three clauses, it almost always has three places to award marks. Answering all three in a single flowing paragraph is not wrong, but it makes the evaluator do the mapping, and an evaluator working through a stack of submissions will not do it generously. Mirror the sub-parts, in the order they were asked, and let the structure of the answer show where each mark lives. This is the cheapest scoring gain available and it costs nothing but discipline.
Where the answers come from
Here is the part the bid writing guides tend to skip, and it sits awkwardly against everything above. Almost none of the raw material in a tender response is new. The capability answer, the methodology, the risk register, the certifications, the reference contracts: the organisation has answered all of it before, often several times, in submissions nobody can now find. That is not an argument for pasting the last submission in. The fourth column of the table above is a catalogue of what happens when you do. It is an argument about where the hours go: they go to hunting for the source material and to checking whether it is still true, and what is left over is the tailoring that actually earns the marks.
That is the problem SEQUESTO's agentic Operating System (aOS) is built for. Approved content lives in the Knowledge Hub and comes back with its source attached, so an answer arrives with the contract, the date and the owner it came from rather than as text of unknown age. The Agent Force, the specialist agents that execute each step, drafts each question against that approved content and keeps a record of what was used, changed and approved. James orchestrates the response from intake to submission and routes each question to the person who owns it. Bid writers keep the argument, the tone and the final word, and the hours the hunting used to take are the hours the tailoring needs. If you want the mechanics of holding evidence this way, the piece on building an auditable content library covers it, and tender response automation is where it runs in practice.
None of this decides whether you should have bid at all. That question belongs earlier, at the go or no-go decision, and the go versus no-go article takes it properly. It is worth being honest about: a rushed response to a strong opportunity scores worse than a resourced response to a smaller number of them. Writing well is also not the whole of winning: the reasons competent bids still lose are structural rather than stylistic, and how to win tenders sets them out. The wider sequence, from opportunity through to standstill, is set out in the tender process guide, and the operational side of running several at once sits under tender management and bid management.
Every response, handled. The final word, yours.
