What actually changed in B2B marketing this year?
Where your buyers go first. They've moved the start of the research journey from a search bar to a chat box, and that quietly rewrites the whole game. More than half of buyers now begin their research in an AI chatbot, per the G2 2026 Buyer Behaviour report, and a majority of Google searches now end with no click at all. If the answer is delivered inside the interface, ranking underneath it stops being the prize. Being the thing the answer cites does.
That's the real shift, and it's not hype. It changes what content is for: less about winning a blue link, more about being the source an answer engine trusts enough to quote. We go deep on the how in AEO, GEO and SEO, so I won't relitigate it here. I want to spend this post on the thing people are getting badly wrong while they chase the shift.
What's the biggest mistake everyone's making?
Automating the one thing that was never meant to be automated: individual human connection. In 2026 the move everyone reached for was letting AI handle personal engagement at both ends of the funnel. On the marketing side, AI writes the thought-leadership posts, then writes the comments on other people's posts, then writes the replies to the comments. On the sales side, it writes the cold emails and the call scripts. It all feels like leverage. It's the opposite.
The whole point of a comment or a one-to-one email is that a person saw you and bothered to respond. Automate it and you keep the format while deleting the meaning. You've industrialised the exact interaction whose only value was that it wasn't industrialised. I've watched teams cheer a jump in "engagement" that was entirely their own bots talking, and wonder later why none of it turned into a conversation.
You automated the one part of marketing whose entire value was that it wasn't automated.
Why can everyone spot AI slop instantly?
Because it has a smell, and by now we've all learned it. The tells are consistent: fluffy and faintly corporate, a bit Deloitte, three or four tidy paragraphs where one line would do, a scattering of em dashes, and a reply that barely engages with the actual post it's replying to. You clock it in half a second and your opinion of the sender drops with it.
The uncomfortable bit is what the slop is optimising for. It isn't optimising for the reader. It's optimising for the algorithm: volume, keywords, an activity graph that looks busy. It games the feed and loses the human, and in B2B, where deals run on trust built over months, that's a genuinely bad trade. You can win the impression and lose the relationship in the same post. I wrote a whole piece on this pattern in where AI makes B2B marketing worse, because it's the most common way I see good teams shoot themselves in the foot.
So what's the right pattern?
AI preps and surfaces, the human connects. Keep the machine on everything up to the moment of connection, and keep a person on the connection itself. That line is the whole discipline.
On social it looks like this: AI finds the posts and people worth engaging with, hands you the context ("they posted this, they recently changed roles, there's an angle here about X"), and then you write the actual reply, in your own voice, because you actually read it. On sales it's the same shape: AI does the research, assembles the account picture, drafts the brief and flags the timing, and the human writes the email and takes the call. You get the leverage of the prep without vandalising the connection. That's the AI-earns-its-place version, and it's the one buyers can't hold against you, and it's the frame behind our whole approach to AI transformation.
What's real and what's hype, in one line?
Real: buyers researching in answer engines, which means GEO and being cited genuinely matter now. Hype dressed as progress: automating comments, replies, cold emails and scripts, which mostly manufactures slop and erodes trust. The rule that survives all of it is boring and correct: hand AI the process, keep the human on the judgement and the relationship. Everything else is noise chasing an algorithm.
If you want an honest read on which of your AI experiments are compounding and which are quietly costing you trust, book a free Growth Audit and we'll go through it with you. No lock-in, and you own everything.
Frequently asked questions
What genuinely changed in B2B marketing in 2026?
Where buyers start. Over half now begin research in an AI chatbot per G2's 2026 report, and most Google searches now end without a click. The prize shifts from ranking under the answer to being the source the answer cites, which is what GEO is about.
Should I use AI to write my LinkedIn comments and cold emails?
No. That's the year's most common mistake. The value of a comment or a one-to-one email is that a person actually engaged, and automating it keeps the format while deleting the meaning. It reads as slop, it games the algorithm rather than the reader, and in trust-based B2B it costs you more than it earns.
How can people tell content is AI-generated?
It has a recognisable smell: fluffy and corporate, three or four tidy paragraphs where one line would do, a scattering of em dashes, and replies that barely engage with the post they're answering. Readers clock it in seconds and trust the sender less for it.
What's the right way to use AI in engagement and sales?
AI preps and surfaces, the human connects. Let AI find the right posts and people, assemble the context and draft the brief, then have a person write the actual reply, email or call in their own voice. You get the leverage of the prep without automating the connection that was the whole point.
