The bid is due in five days, three technical sections are unanswered and the clarification window has closed. How do you run the last five days?

Marks, not discomfort

With five days left you have a fixed capacity and a variable return, so the first hour goes on arithmetic rather than writing. Most public and semi-public procurements publish the scoring model, and where they do, the allocation of your remaining effort is a calculation rather than a judgement.

Assume a technical evaluation worth 300 marks across fourteen requirement groups, and roughly six usable person-days of senior capacity left after allowing for review and submission.

Work available	Marks at stake	Realistic gain	Marks per person-day
Section 6, unanswered, resilience and DR	45	0 to 30	~15
Section 9, unanswered, data migration	30	0 to 20	~13
Section 11, unanswered, service management	20	0 to 12	~8
Compliance matrix, half populated	gates the whole score	pass or fail	not comparable
Rewriting section 3, already answered, scored well	40	2 to 4	~2

The last row is where bids die. Section 3 is the one the team wants to polish because it is the interesting part and it is nearly good, at two marks a person-day the four marks of uplift cost two days, and those same two days would have banked the full twenty on migration with half a day left over. An unanswered requirement scores zero regardless of how strong the rest of the document is, and closing zeros is the only work with a genuinely high return this late.

The compliance matrix sits outside the comparison because it is not a scored section, it is the interface to scoring. A half-populated matrix risks the whole submission being marked non-compliant or having rows scored against whatever the evaluator can find, so it gets finished whatever else does not.

Freeze the solution and the price on day one

The characteristic defect of a bid finished under pressure is not weak prose, it is internal contradiction: the architecture section describes two availability zones, the DR answer written on day four describes three, and the price was built on one. Evaluators find these, and a document that disagrees with itself invites every claim in it to be discounted.

So day one ends with two artefacts frozen and circulated: a one-page solution statement naming the architecture, the deployment model, the numbers you are committing to and the phasing, and the price with its assumptions. Everything written afterwards is written against that page. If the solution genuinely has to change after the freeze, the change is made by one named person who then re-reads every section it touches — which is expensive, and knowing it is expensive is what stops people making the change casually.

Answering a section you lack the information for

The unanswered sections are usually unanswered because nobody knows the answer, and the clarification window has closed. Inventing a figure here is the one unrecoverable move, because a claim that fails in due diligence removes the credibility of every other answer and occasionally removes you from the process.

What works instead is to answer the requirement with the part you can commit to, state the boundary explicitly, and make the unknown a stated assumption with a consequence attached.

"We commit to a recovery time objective of four hours for the claims intake service, measured from declaration of a disaster, on the basis that the warehouse and the identity provider are recovered by you within the same window. Our answer assumes the DR environment described in your section 2.4 is sized to production capacity; if it is provisioned at reduced capacity, the objective becomes eight hours and we will confirm this at contract award. We have not been able to verify this point within the clarification process and have stated the assumption rather than assume in our own favour."

An evaluator reading that sees a commitment, a boundary and a candid statement of what is unknown. It will not score as well as a full unqualified compliance, and it scores far better than a paragraph of confident prose that a technical panel can tell was written without knowledge of their estate.

The last day is not for writing

Reserve a full day, and protect it. It covers three things that each independently lose bids: a red-team read by somebody who has not written any of it, checking that the compliance matrix references resolve to real page numbers, and the submission mechanics themselves.

Mechanics deserve the same seriousness as content. Portal uploads have file size and format limits, some require each section as a separate document, signatures and certificates are frequently mandatory attachments, and portals close on a hard clock that does not care that your file was rendering. Submitting six hours early with a slightly weaker section 11 is strictly better than submitting a stronger document at the deadline, because one of those outcomes has a failure mode where you score nothing at all.

The five days were decided in week one

The interviewer is also listening for whether you know how this situation is manufactured. It comes from starting with the solution and the interesting sections, leaving the compliance matrix as an administrative task for later, and not identifying on day one which requirements need somebody who is not on the bid team. A bid run properly is sequenced in the opposite order: matrix first, so every requirement has an owner and a deadline; the sections needing external input commissioned immediately, because their lead time is other people's calendars; and the narrative written last, when the answers it has to summarise exist.

Saying that unprompted is what distinguishes a candidate who has run bids from one who has contributed to them. The second one describes heroics in the last week. The first one describes a plan whose last week is a review.