Your best solutions engineer can explain what your product does, and why it beats the alternative, better than anyone on your marketing team. You asked her to write one blog post about it. You asked in March. It’s June, the draft doesn’t exist, and you both feel vaguely guilty every time it comes up.
This is the subject matter expert content problem, and almost every B2B technical company has a version of it. The instinct is to read it as a discipline problem: the expert won’t prioritise writing. Or a capacity problem: we need another writer. Both readings are wrong, and acting on either one makes the content worse.
Buyers can tell the difference, and they’re keeping score
Start with why this content matters more than the rest of your output. When Edelman and LinkedIn surveyed business decision-makers for their 2024 report, 73% said an organisation’s thought leadership is a more trustworthy basis for judging its capabilities than its marketing materials and product sheets. Ninety percent said they’d be more receptive to a sales or marketing approach from a company that consistently produces high-quality thought leadership.
So the expert-grounded piece you can’t get written is precisely the piece your buyers weight most heavily. Here’s the part that should sting: the same research found that good is no longer good enough. Most decision-makers rate the quality of what they actually read as mediocre, and companies openly admit they don’t put their most senior and talented people into the work.
Most companies respond to that gap by publishing more. The buyers reading it are already unimpressed by the volume that exists.
The bottleneck is expert time, not writing capacity
Ask content teams what slows them down and the same answers come back every year. In CMI’s 2025 research, resource constraints, time, people, and budget ranked among the top challenges, and only about one in three B2B marketers said they had a scalable model for producing content at all. Roughly a third named accessing SMEs specifically as a problem.
That last number is the one worth sitting with. The blocker isn’t a shortage of people who can write sentences. It’s a shortage of access to the people who know which sentences are true.
It’s tempting to blame the experts. Some genuinely resist the process, and that’s real. But the dominant pattern that experienced content leaders describe is simpler and less personal: content isn’t on the expert’s list of objectives, so it loses every time it competes with the work they’re actually measured on. The architect isn’t refusing to write. She’s shipping the thing you hired her to ship.
The teams who solve this stop asking experts to author and start treating them as a source. The model they converge on is structured extraction: a writer does the research first, then runs a focused 30-to-45-minute interview that pulls out the specifics, the caveats, and the one thing the expert thinks the rest of the category gets wrong. One good session feeds several pieces. The expert’s calendar barely notices.
Why hiring another writer makes it worse
The obvious fix is more writing capacity. It’s also the trap. A skilled generalist who doesn’t understand your product writes content that reads fluently and says nothing a technical buyer respects. The prose is clean. The substance is hollow, and your buyers are exactly the people equipped to notice.
You might be thinking a good writer can simply research the topic. They can, up to the surface. The depth that makes technical content credible, the edge case, the “it depends, and here’s what it depends on,” the caveat a vendor’s own docs leave out, lives in the expert’s head and is not sitting on page one of a search result. Content leaders who’ve tried to scale with generalist writers report the same failure mode: talented people, competent copy, output too generic to be worth publishing, and a quiet decision to bring the work back in-house.
More capacity doesn’t clear the bottleneck. It just produces more of the content buyers already discount, faster.
AI raised the floor and the bar at the same time
This used to be an expensive problem. Now it’s a sharper one, because anyone can generate fluent, well-structured, plausible content in minutes, and most of the internet is busy doing exactly that.
That changes the maths in a way worth understanding. An Ahrefs study that ran 600,000 top-ranking pages through an AI detector found that 86.5% showed some level of AI assistance, and the correlation between how much AI a page used and where it ranked was 0.011, which is statistically meaningless. Production method, on its own, predicts nothing. The same study found something more pointed underneath that headline: pages built almost entirely from AI rarely reached the top spot, and the highest-ranking results skewed toward lighter AI use and human input. The deciding factor wasn’t who or what typed the words. It was whether the page said something the other results didn’t. Google’s position has been consistent through its 2026 core updates: it filters low-quality, thin, mass-produced content regardless of how it was made, and rewards content that demonstrates real experience and expertise.
Read those two facts together and the conclusion is hard to dodge. AI has made generic content free. Free generic content is worthless content. The only input that’s still scarce, and still defensible, is verified expert knowledge. The bottleneck you’ve been treating as an annoyance is now the whole game.
What actually clears the bottleneck
Reframe the expert’s job from author to source, and build the process around protecting the part only they can do.
A version that works in practice: do the research before you ever take the expert’s time, so you arrive able to ask sharp questions instead of “tell me about onboarding.” Run one structured 30-to-45-minute interview, or a tight asynchronous questionnaire for someone who’d rather write than talk, and pull out the specifics, the trade-offs, and the contrarian point. Hand that to a writer, or an AI-assisted draft, to shape into a structured piece. Send it back to the expert for an accuracy review only, which takes minutes because reviewing is faster than creating from a blank page. Then verify every external claim before it ships, because a technical reader will check, and being confidently wrong costs more than saying nothing.
This is also where AI earns its place, and where it doesn’t. It compresses the research and the drafting, which is most of the calendar time. It cannot supply the expert’s judgment, and it cannot tell you whether a claim is actually true. So the human stays exactly where the value is: the expert’s specifics going in, real verification coming out.
The question to put to your team this week isn’t “who can write this?” It’s “whose knowledge does this piece need, and what’s the smallest slice of their time that gets it onto the page?” If the honest answer is still three days of the architect’s calendar, the post will not exist in June either. Build the process so the answer is forty-five minutes, and it will.