Noticed it? Everybody’s publishing at full tilt but it all sounds a little dead inside.
No real surprise. Salesforce found 71% of marketers already use generative AI in their workflows. Adobe reckons 81% of workers say generative AI is fundamentally changing how marketing work gets done.
It all moved stupidly fast, too. From experimental toy to baked-in marketing workflow in the space of five minutes.
You can see it everywhere. The blogs ship faster and the nurture sequences get longer. LinkedIn (once merely an unbearable corner of professional Hell, now just the same but much, much worse) is a wall of regurgitated hot takes that all sound assembled by the same haunted content demon that lives inside ChatGPT.
Volume won. We’re all scaling output. Everybody’s building ‘thought leadership engines’. And underneath all the workflow diagrams and agentic workflow-of the-future enthusiasm, loads of scaling B2B companies are dealing with the same problem.
Everything’s starting to sound the fucking same.
Most AI-assisted B2B content isn’t terrible. That’s the weird (and annoying) part.
It’s clear enough, structured enough, optimised enough. The grammar’s fine. The SEO boxes are ticked and the CTA’s in roughly the right place. And yet sometimes it just never really lands.
Read three SaaS landing pages in a row these days, and your brain starts leaking out your ears. The tone converges. The phrasing converges. Even the ‘bold POV’ thought leadership starts sounding suspiciously similar once you’ve seen enough of it. Everybody’s differentiating in exactly the same voice.
Meanwhile buyers are consuming more content than ever before. Gartner research suggests B2B buyers now spend only a tiny fraction of their buying journey actually talking to suppliers directly. Most of the process happens independently through research and peer validation and digital content.
That means your content is doing more of the selling than ever. Problem is, buyers are getting weirdly good at spotting vanilla.
Most buyers won’t say:
“Ah yes, this paragraph was likely assembled using probabilistic language generation models.”
They’ll just kill the page and go about their day. Because people can feel when nobody really meant what they’re reading. That’s the bit a lot of AI chatter misses. The issue isn’t accuracy. AI is (quite) often technically accurate.
The issue is more a kind of emotional flatness. There’s no texture, no specificity, not enough of that ‘lived-in’ quality or the sense that somebody with skin in the game sat down and tried to communicate something real.
Everything starts sounding polished, sensible and completely interchangeable. The uncanny valley, it seems, is now littered with whitepapers.
Edelman’s Trust Barometer found only 32% of people trust AI-generated information. That’s not a big number. And it means most buyers (your buyers) don’t remember technically competent content, but they do remember the stuff that sounds human enough to survive the slop flood.
A few years ago, the bottleneck was production. You literally couldn't make the stuff fast enough. That problem is now solved. Congrats, we fixed it!
The new bottleneck is perspective. Specificity, credibility, taste, the discerning eye, the narrative instinct. Knowing when something is bland. Knowing when a sentence sounds like bullshit.
That's editorial judgement for you. And it's exactly the thing AI still can't fake convincingly. The skilled editor just levelled-up to be the AI vanilla killer - and If nobody's playing that role, vanilla is winning...
Read more about what it could cost if your content all sounds the same.
This isn't a framework or a methodology. It's just five questions that separate content worth reading from content that just ships.
If the answer's no, it's not a point of view, it's a vibe. Rewrite it until someone might feel so inspired/happy/incredulous/aroused/outraged that they click your CTA.
A specific client situation. A real observation. Something that doesn't survive being copy-pasted to a competitor's blog. If not, find it and fix it (or kill it).
Read it out loud. If it sounds like a Terms and Conditions page with misplaced ambition, it needs another hard, human edit.
If you can't answer in one sentence, the content can't either.
I don't mean publish it. Would you send it to your old boss or your latest crush? If you're not sure, you know what to do.
None of this requires scrapping the AI workflow. But it deffo does need someone in the room who gives enough of a shit to ask the questions.
They're the ones with editorial control baked in. A real point of view, consistently applied. The ones with content that's specific enough to be credible and has a turn of phrase that makes your cynical buyer stop scrolling for even half a second.
In short, they're teams who already have their very own vanilla killer.
It's a totally different skillset from production, and it's a skillset that's arguably more important now than it's ever been before.