The content job is moving up a level

I went to Accountex North this week with a fairly specific list of things I wanted to understand. It wasn’t a list of articles I wanted to write. It was a list of questions: what are accountants actually misunderstanding about Making Tax Digital? Where is the extra operational work appearing? What changes between a quarterly update and a Final Declaration? What happens when a client has multiple income sources? And, underneath all of that, where are the evidence gaps in the content we already have?

That changed how I covered the event. I chose sessions because they might answer those questions, recorded and transcribed what I heard, and started comparing what HMRC, software providers and practitioners were saying as the day went on.

Some things repeated. Some complicated the assumptions I’d arrived with. Some were useful enough to preserve even though I don’t yet know what they’ll become. By the end of the day, I’d accumulated something more useful than notes for an event recap: I’d built a small evidence bank from six sessions.

The output came later

This sounds like a minor distinction, but I don’t think it is. For most of my career, content work has tended to start with an output: write the article, interview the customer, cover the event or build the campaign page. The output gives the work its shape.

At Accountex, I found myself doing almost the reverse. I started with what we didn’t know, gathered evidence against it, looked for patterns and only then started asking what the material might become.

There are obvious outputs in it: an article, FAQs, video, material designed to answer specific questions in search and AI systems, or internal intelligence for people working on the same subject. There are also things that may never become published content, and that’s fine. The evidence still has value if I can find it again when the next question appears.

I’ve written before about making work recoverable and about how prompting is becoming routing. This feels related to both. The useful unit of work isn’t always the finished asset anymore; sometimes it’s the evidence, decision or context that makes a future asset better.

Pipeline is the end of the chain

After returning from Accountex, I read a State of Brand article about companies hiring editors to run thought leadership and measuring them against pipeline. The example that caught my attention was Workato, where thought leadership sits closer to sales than communications and is judged using measures such as win rates, deal velocity and average sales price rather than simply reach and downloads.

It is only a small hiring signal, and I would be wary of declaring an entirely new profession from a handful of job descriptions. But it extends something I’ve been tracking in the editorial jobs are moving upstream: organisations are asking editorial people to own narrative, evidence, systems and orchestration rather than simply a production queue.

The commercial accountability matters. Content teams have spent years reporting what was easiest to count: articles published, page views, downloads and leads. Those measures can show activity without showing whether the work affected how anyone understood the market or evaluated the company. If thought leadership is supposed to influence a buying decision, it shouldn’t be permanently protected from questions about whether it did.

But pipeline is the end of the chain, not the beginning. Before someone enters a sales process, they may search for an answer, ask an AI system, read an expert, encounter a useful argument or recognise their own problem in evidence a company has published. A colleague the sales team never meets may use that material to shape the decision internally. The eventual opportunity can appear in a CRM; most of what prepared the opportunity will not.

There is a real risk here. If pipeline becomes the only acceptable measure, the senior editor doesn’t necessarily gain strategic authority. They may simply become a demand-generation manager with a more editorial job title. Short-term attribution could pull the work towards topics closest to an immediate sale, even when the more valuable opportunity is to build authority around a problem the market hasn’t named clearly yet.

The answer isn’t to avoid commercial measurement. It is to make the chain leading to it more visible: understand the questions buyers are asking, identify where the organisation lacks credible evidence, develop a useful point of view, make it retrievable and watch what happens after publication. Pipeline belongs in that system, but it can’t explain the whole system by itself.

GEO changes where the work starts

This is where generative engine optimisation becomes more interesting to me than simply adjusting published copy for AI systems.

The GEO work behind my Accountex visit had already surfaced questions around quarterly updates, Final Declaration, client readiness, record keeping, multiple income sources and operational workload. I didn’t treat those questions as a list of articles to manufacture. I used them as a reporting brief.

That distinction mattered because the evidence changed the emerging argument. The broad assumption was that more frequent reporting automatically creates more work. What I heard across the day suggested that the pressure sits less in the act of submitting an update and more in the operating system around it: transition, client readiness, chasing, workflow design, exception handling and work compressed into deadline periods.

That creates a more specific and defensible bridge between Making Tax Digital, capacity and hidden administrative work. It can also produce several useful forms: a direct answer explaining quarterly updates and Final Declaration, practitioner-led guidance on client readiness, video responses, page improvements and material for a wider narrative about where invisible work is created.

The demand signal helped identify what needed answering. Editorial judgement was still required to decide what evidence was missing, which sources to trust and whether the original framing survived contact with practitioners.

GEO can therefore influence work before production begins. It can help turn search and AI behaviour into editorial intelligence, then make the resulting evidence easier for people and machines to retrieve. That’s the larger opportunity I’ve been exploring in becoming the GEO journalist: not simply learning a new optimisation technique, but using visibility questions to decide what evidence is missing and then going out to find it.

It shouldn’t determine the entire editorial agenda. Some valuable ideas begin with observation rather than recorded demand. One of the other themes I brought back from Accountex started that way: a practical argument about boundaries and recovering capacity. The next step is to see whether it can be made answerable and useful, not to discard it because it didn’t begin in a prompt report.

Demand can tell us what people are asking. Editorial judgement still has to notice what might matter next.

The evidence should survive the asset

Accountex made this pattern particularly visible, but it isn’t the first place I’ve noticed it. I’ve been experimenting with customer stories in a similar way. Traditionally, you interview a customer, write the story, publish it and move on.

But the interview contains much more than the published case study. There are proof points, useful phrases, unanswered questions, themes that might recur elsewhere and questions you wish you’d asked. So I’ve been testing what happens if some of that intelligence survives the finished story. The case study remains an output, but it also leaves something behind for the next piece of work.

That’s closely related to the idea that content teams need an evidence layer: finished content shouldn’t be the only thing the work leaves behind. I’ve noticed the same thing with narratives.

An idea doesn’t necessarily stop at the campaign that introduced it. Language travels. Other people adapt it. A phrase changes while the underlying argument survives. That makes watching what happens after publication part of the editorial job too: what did people adopt, what evidence did they reuse, what language travelled and what should the response change next?

This is also where a commercial scorecard could become more credible. A page change, a citation in an AI answer, material reused by sales, a stakeholder adopting the argument, stronger engagement from the intended audience and eventual pipeline influence are different kinds of evidence. None proves the whole case alone. Together, they can show whether the point of view moved.

AI makes the surrounding system more important

There’s a popular version of the AI content argument that says one person can now become a writer, designer, videographer, strategist and distribution team at the same time. I’m not sure that’s the most interesting change. It risks turning AI into an argument for making one person do several jobs.

What I’m finding more useful is that AI makes it easier to work with the material surrounding an output. I can take several transcripts and look for recurring themes, interrogate notes from different sessions, turn evidence into different forms, preserve a decision and recover its context weeks later, or ask what two apparently separate projects have in common.

Production gets easier, but that puts more weight on the decisions around production. Which questions were worth asking? Which source should I trust? Is this actually a pattern or did two people happen to use similar language? Do we have enough evidence to make the argument? Is another article useful, or are we publishing because an article was in the plan?

AI can help me work through those decisions. I still have to own them. It can also make an argument easier to find, extract and reuse, but retrievability is not a substitute for having something useful and well-supported to retrieve.

Maybe this is what seniority starts to look like

Content careers have often had an awkward definition of progression. You get good at making things, then seniority gradually takes you away from making them: more management, more coordination, more meetings and more distance from the actual material.

But there may be another kind of progression. You can stay close to the work while owning a larger part of the system around it: identifying the information gap that makes an article necessary, improving what the organisation remembers from interviews, deciding what evidence is worth bringing back from an event, making the resulting knowledge retrievable and watching whether a narrative travels.

That’s different from becoming a full-stack content creator. If anything, it’s less about doing every production job yourself and more about taking responsibility for the editorial decisions connecting them. I’ve started thinking of that as editorial ownership.

The new thought leadership roles measured against pipeline may be one organisational expression of it. They suggest that at least some companies want editors to own more than a publishing calendar. But accountability needs to be matched with authority. If editors are going to be measured against pipeline, they need some ownership of everything that happens before it: the questions buyers ask, the evidence the company holds, the point of view it develops and whether that thinking can be retrieved when a decision is being shaped.

I can’t claim that my Accountex work has influenced pipeline. It hasn’t yet produced a published page change or an improvement in AI visibility. The fieldwork and first synthesis are complete; adoption and performance remain open. That distinction matters because building a more traceable editorial system isn’t the same as proving its commercial effect, although it gives us a better chance of doing so.

I still like writing. I still think interviewing someone properly matters, and I still care about the quality of the finished thing. Increasingly, though, the finished thing feels like one part of the work rather than its boundary.

The job is moving outwards: into the questions before the content, the evidence underneath it, the systems that make it retrievable, the commercial conversations it supports and the memory left behind afterwards. If AI keeps making production easier, that surrounding layer of judgement may be where more of the value ends up.