Most of the article was decided before I wrote it

The work behind Beyond the brief, Sage’s companion article to the first Hidden Hours of Accountancy episode.

This week Sage published Beyond the brief: Where do accountants’ hidden hours go?, which I wrote as the companion to the first episode of The Hidden Hours of Accountancy video series. It opens with a practice owner, Andrew Dick of G W Dick & Co, saying that with five extra hours a week he’d “probably stop working on a Saturday”. It draws on Sage’s research with 1,000 UK accountants and bookkeepers, the episode’s conversations with Andrew and with Julie Young of Walter Dawson & Son Accountants, and Aaron Patrick’s account at Accountex Manchester of how the first MTD for Income Tax quarterly update forced him to rethink his capacity planning.

Drafting still required choices about emphasis, structure and language. But the questions that made those choices possible were largely settled earlier: what job it had, which questions it should answer, what counted as useful evidence and which voices could carry the argument. I’ve written separately about the campaign-level groundwork behind Hidden Hours. This piece is narrower: how one article came together, and what that says about where editorial work now sits.

The article had a job before it had a draft

The four Hidden Hours episodes were planned as an editorial and search ecosystem rather than four standalone films. Each episode was treated as one evidence-led editorial hypothesis, then checked against search and GEO analysis for terminology, demand and overlap with existing Sage content. The first episode, about how practitioners actually spend their time, was linked to questions about non-billable work and where the working week goes.

That gave the article a defined role. It wasn’t a recap of the video or another summary of the research. It had to answer a practical question, where the work around the work comes from and what a practice can do about it, in a form that stands on its own for someone who never presses play. Knowing that in advance ruled out a lot of plausible articles.

I had also drafted a small diagnostic set of AI-visibility questions for the episodes, such as how much work accountants do outside their agreed scope and how firms can reduce scope creep. It was never meant as a permanent prompt library, only a sample of the territory. Its influence shows in the article’s FAQs, which answer how to spot scope creep early and whether access to a client’s software means the accountant is monitoring it.

Deciding what counted as evidence

Before the shoot, I wrote an optional prompt pack to sit alongside the episode brief. It was not a replacement interview guide, and the episode had its own brief, participants and production team. The pack’s purpose was narrower: to make it more likely that the conversations produced specific, quotable, first-party evidence rather than general opinion.

It translated search and AI-prompt themes such as productivity, managing many clients, automation and practice capacity into practitioner questions, and deliberately avoided software-buying questions. Instead it asked what takes up far more of the week than outsiders would expect, what sounds like a quick request but turns into a much bigger job, and where technology saves time rather than simply moving the work somewhere else. It also set a bar for a useful answer: a specific example, a before-and-after comparison, an estimate of time, a consequence for capacity or personal time, or a concrete change that helped.

The pack proposed three tests for the resulting footage. Is there a recognisable human story? Is there concrete evidence? Could the answer help someone asking a real question in search or an AI tool? The order matters, because it keeps GEO downstream of good journalism. Prompt data could suggest what was worth asking about, but it shouldn’t dictate the conversation.

That standard is visible in what made the article. Andrew’s Saturday and Julie’s wish for time to plan the future of her own practice are specific, personal and consequential, and they turn a statistic about switching off into something a reader recognises. One of the pack’s closing questions was about getting five hours back, which is the question the article is built around. I’d be overstating it to claim the pack produced those answers on its own. What it did was describe in advance what a useful answer would look like, so I knew one when I saw it.

Some of the best evidence wasn’t planned

The strongest counterargument to all this preparation is that the best material often arrives unannounced, and that is partly true here. Aaron Patrick’s contribution didn’t come from a planned interview question. It came from a conference session at Accountex Manchester that I was reporting on.

What preparation changed was what I listened for and what I kept. My Accountex plan started from evidence gaps rather than event coverage: decide what Sage needed to know, then use access to practitioners to fill or test those gaps. When Aaron described his first MTD quarter, I noted that filing wasn’t his main difficulty. Engagement letters, client expectations, getting data in on time and capacity were. I recorded how he planned to respond, by setting expectations about fees and requirements much earlier, and his embedded-accounting example, alongside guardrails: separate his experience from product claims and HMRC guidance, and verify quotations before publishing.

A month later that note became one of the article’s most useful sections. The planning didn’t create the evidence, but it made it recognisable when it appeared and kept it usable after the event. That is a more modest claim than “plan your evidence in advance”, and I think a more accurate one.

Decisions the reader never sees

Several choices shaped the article without announcing themselves. There is no product pitch in the body. The research and the series carry the Sage association, and the article’s job was to be useful enough that a practitioner would trust where it came from. Aaron’s embedded-accounting example involves a banking app rather than Sage, because that is what he described, and the article presents it as a way to judge a connected workflow rather than as a recommendation.

The five-hours question became the frame because it turns capacity into a personal stake. The article opens with Andrew’s answer and returns to both answers near the end, where they show two things hidden hours displace: rest, and time to plan. The practical section is four steps, from recognising recurring support to making room for judgement, and it ends on one action a reader can take with their own last working week. Those were structural decisions, and they were easy to make because the material was already specific.

What this does and doesn’t show

The article is a working example of something I’ve argued on this site: that editorial work increasingly happens before the brief, in deciding questions, evidence standards and destinations. By the time I was drafting, much of the direction had been established. Writing still meant choosing what to emphasise, how to connect the evidence and what a reader could reasonably take from it. The point isn’t that writing stopped requiring judgement. It’s that the judgement began long before the first paragraph.

It doesn’t show impact. The four-video plan proposed recording how Sage is represented for the question set before activation and checking again afterwards. Until that comparison exists, I can’t say the article improved anything in search or AI answers, and I shouldn’t. Nor do I own the wider programme: the video series, its participants and its production belong to the wider team. My contribution here was the question and evidence design around one episode, the event reporting and the article itself.

The next useful step is measurement, not more content. If the question set shows Sage’s practitioner evidence being cited where generic answers used to appear, the approach has earned a wider test. If it doesn’t, that is worth knowing too.