Draft Fast, Check Hard

A communications team on state and federal rounds lives between two failure modes. The blank page, which costs hours the news cycle does not give back. And the wrong draft — the media release with the mangled figure, the talking points that promise what the policy does not — which costs credibility that no cycle gives back. Assistants have made the first problem cheap to solve. Solving it without inflaming the second is a matter of process, and the process fits on a page.

Codify the voice before you need it

The single highest-leverage document a comms team can write is a voice-and-language reference: the words we use, the words we never use, how we describe the people we work for and the people we disagree with, the sentence rhythm we favour, the clichés we have banned. Most teams carry all of this in the heads of two senior people, which is why every draft needs those two people. Written down once, it becomes something an assistant can be loaded with — and something a new staffer can learn in an afternoon instead of a year. It is the difference between a tool that writes like the internet and a tool that writes like you.

The first draft is an argument, not an answer

With the reference loaded and the facts of the announcement supplied, a competent first draft of a media release takes a minute. The failure is stopping there. Generic copy is what accepting the first pass produces; the craft is in the iteration — push back on the weak lead, ask for three sharper versions of it, cut the third paragraph and see what the piece loses, tighten the quote until it sounds like something a person would actually say aloud. Treat the assistant as a fast, thick-skinned junior writer whose work you edit hard, and the drafts improve run over run. Treat it as an oracle and you will publish its mediocrity under your name.

Separate the claims from the prose

When the draft reads well, stop reading it as prose. The fact-check pass begins by extracting every checkable claim into its own list — every figure, every date, every attribution, every characterisation of what a policy does or an opponent said. The extraction itself is mechanical work an assistant does well. The verification is not. A person checks each consequential claim against the primary source: the budget paper, the Hansard record, the published report, the actual transcript. Not against a summary of the source, and never against the model's own confidence.

Then one last read, by a person, purely for voice — ideally aloud. Accuracy checks catch what is false; the voice read catches what is true but not yours: the borrowed phrase, the register that drifts formal, the sentence no one on your team would write. It takes five minutes and it is the difference between material that sounds issued and material that sounds meant.

One briefing, many rooms

The pipeline's real return arrives at the repurposing stage. A recorded briefing, a roundtable, a long policy conversation — transcribed, it becomes source material, and the same verified substance can be recut for every room it needs to reach. The internal memo for the leadership team. The opinion piece, pitched to a metro masthead under the principal's byline, making one argument well. The week of social content, each post shaped to its platform rather than pasted across all of them. None of these starts from a blank page, because the thinking was done once, in the meeting, and captured.

Each format still gets its own review, because each carries its own risks — the memo can be frank where the opinion piece is legally exposed, and a claim safe in eight hundred words can be defamatory shorn to eighty. The efficiency is real, but it lives inside the discipline, not instead of it. Draft fast. Check hard. In that order, every time.