The clock that actually matters

A state government panel tender drops on a Tuesday. Closing date is fourteen days out, 2pm, no extensions. Sixty-odd questions across six schedules: capability, methodology, key personnel, local content, WHS, pricing. Two of your subject-matter experts are on client site until the following Thursday. The bid manager has three other live submissions.

Everybody in that room knows how the fortnight goes. Days one to three are triage: read the pack, build the compliance matrix, chase the addenda, argue about whether to bid at all. Days four to nine are collection, which mostly means sending people questions and waiting. The first genuinely complete draft appears somewhere on day ten or eleven. Review, sign-off, formatting, and upload eat the rest.

So the draft that decides the outcome gets, at best, three days of improvement. And those three days are not spent on strategy. They are spent finding out that answer 4.2 contradicts answer 2.7, that nobody addressed the sub-question about subcontractor management, and that the methodology section reads like it was written by three different firms, because it was.

The metric that governs tender quality is not writing speed. It is time-to-first-draft: the number of days between the tender landing and the moment a complete, compliant, end-to-end draft exists on one screen. Everything good happens after that moment. Every hour before it is overhead.

Why the first draft arrives so late

It is tempting to blame the SMEs. That is almost never the real cause. Four structural things push the first draft to the back of the timeline.

1. The response is a list, and lists get treated as prose

A tender is not one document. It is sixty small documents with a cover on top. Each question has its own word limit, its own evaluation weighting, its own mandatory inclusions. Teams that treat the response as a narrative to be written top-to-bottom spend the first week fighting the shape of the thing rather than answering it. We wrote about this failure mode in repeating sections: list-shaped content drafted as prose is where items get dropped, merged, or quietly invented.

2. Most of the answers already exist, in the wrong shape

Your capability statement answers question 1.1. It just answers it at 900 words for a different client, with a different emphasis, from an eighteen-month-old submission. The WHS material exists in your ISO documentation. The key personnel content exists in four CVs and a capability matrix. Nothing is missing. Everything needs re-cutting, and re-cutting by hand is slower than writing fresh, which is why people avoid starting.

3. Nobody knows what is missing until review

The gaps in a tender response are invisible while it is in pieces. Question 3.4 asks for a transition plan including a named transition manager and a risk register reference. You have the plan. You do not have the named manager, because the person who knows has been on site all week. That gap does not surface until the first consolidated read, on day eleven, when it is expensive.

4. Six authors produce six voices

Distributed drafting is the only way to hit a fourteen-day deadline with sixty questions. It is also the reason the consolidated draft needs a full editorial pass before anyone can evaluate whether the content is any good. The bid manager becomes a copy editor at exactly the point they should be a strategist.

What compresses, and what shouldn't

The useful move is not to write faster. It is to move the first complete draft from day eleven to day three, then spend the recovered week on the things evaluators actually score: win themes, evidence, differentiation, and pricing narrative.

That compression comes from four specific mechanics, none of which is a chat window.

Structure captured before the tender arrives. A tender template that already knows your firm's response shape (the sections, the guidelines for each, which are must-have versus nice-to-have, what information each one requires) means day one is mapping the client's schedule onto a structure that exists, not inventing one. The template layer is the part that carries across every tender you respond to, so the setup cost is paid once, not per bid.

Sources read, not remembered. The RFP pack, the addenda, your last three submissions, the ISO evidence, the transcripts from the go/no-go session: these are the material the draft is built from. When a drafting system reads them directly as sources for this specific document, the re-cutting problem stops being manual. The prior submission becomes input rather than a file somebody is supposed to remember to open.

Questions surfaced early, not at review. This is the single biggest lever, and the least obvious. If the system drafting question 3.4 does not have a named transition manager, the correct behaviour is to stop and ask, on day two, while there is time to get an answer. The wrong behaviour is to write a fluent paragraph around the gap that survives to day eleven and then has to be caught by a human reading carefully. That is the whole argument for a clarifications loop, and it matters more in tenders than anywhere else, because a missed mandatory inclusion is not a quality problem. It is a non-conformance.

One voice, applied automatically. If the firm's voice, glossary, and author identity are workspace assets rather than instructions somebody pastes into a prompt, the consolidated draft does not need an editorial pass to sound like one firm. It already does. That alone removes a day.

The same tender, run differently

Take the panel tender above. Fourteen days, sixty questions, six schedules. Here is what the front half looks like when the structure and sources do the heavy lifting.

Day 1, morning: compliance matrix built from the schedule. The tender template is instantiated for this bid: sections mapped to the client's question numbers, word limits recorded as targets, criticality set per question. Go/no-go call happens against a real structure instead of a hunch.
Day 1, afternoon: sources loaded. RFP pack, two prior submissions from the same buyer, the WHS and quality evidence, four CVs, the pricing assumptions memo.
Day 2: section-by-section generation begins. Each answer drafts against the sources, then waits for a human before the next one starts. Fourteen questions produce clarification requests: missing transition manager, an ambiguity about whether local content is measured by spend or headcount, three claims the sources do not support. Those fourteen questions go out to the SMEs on day two, not day eleven.
Day 3: answers come back. Sections regenerate with the new context, and because each section sees the ones before it, the corrections propagate forward instead of needing to be re-applied by hand. A complete, compliant, single-voice draft exists by end of day.
Days 4 to 12: the part that wins tenders. Win themes sharpened against the evaluation criteria. Evidence strengthened where the score weighting is highest. The methodology section rewritten twice because it is worth twenty percent. A red team review with actual time to act on what it finds.

Nothing in that sequence removed a human from a decision. The SMEs still supplied the facts. The bid manager still owned the strategy. What changed is that the mechanical work happened in two days instead of nine, and the questions were asked while answers were still cheap.

What stays slow, on purpose

Some parts of a tender should not compress, and a tool that claims otherwise is selling you risk.

Pricing is a commercial decision, not a drafting task. Named personnel commitments are contractual. Anything you assert about past performance has to be true and evidenced, which means a human who was there confirms it. Local content percentages, certification numbers, insurance limits, and referee details are facts that live in systems of record and belong in the draft by transcription, not generation.

The distinction is worth drawing sharply, because it is where bid teams get burned. We covered the boundary in more detail in what to standardise and what to keep human. The short version: automate the shape, the assembly, and the first pass. Never automate the commitment.

Measure it before you change anything

If you want to know whether this is worth attacking in your team, take your last four submissions and record two dates for each: the day the tender landed, and the day a complete end-to-end draft first existed. Not a partial. Not a draft with three sections marked TBC. Complete.

Then record what happened after that date. How many days of actual improvement did the draft get? For most bid teams the honest answer is two or three, out of a fourteen-day window. That ratio, not word count and not writing speed, is the thing worth moving. A team that gets to first draft on day three has four times the improvement window of a team that gets there on day eleven, with the same headcount and the same deadline.

It also changes who can run a bid. When the structure, criticality, and required information live in the template rather than in the bid manager's head, a second person can run a parallel submission without the quality falling off. For most boutique firms that is the actual constraint on how many tenders they can chase.

Where to start

Pick the tender type you respond to most often. Capture it once as a template: sections, guidelines, criticality, required information. Load your two best prior submissions as sources. Then run your next live tender through it and record the two dates again.

You will know inside one bid cycle whether it worked, because the first complete draft either arrived early or it didn't. See the tender and RFP use cases for how other teams have set this up, or read the full how-it-works walkthrough for the mechanics.