Free report: “5 Things $39.6M in B2B Programmatic Spend Taught Us.” Three years of managed spend and a billion tracked page loads, turned into five findings platforms won’t put in a pitch deck. Subscribe free to unlock it, plus every new issue of The B2B Stack.
Join 1,600+ B2B marketers, including readers at Dentsu, HubSpot, 6sense, Informa and DirecTV.
What you’ll learn in this article:
Why AI-mediated research has become the most closed, least instrumentable version of the dark funnel that’s ever existed - and frankly there are no fast ‘fixes’
Why buying an ad slot inside the model itself is live in some platforms, reversed in another, and absent entirely in Claude, and why that’s the wrong place to focus anyway
How omnichannel paid media builds the muscle memory that gets your brand typed directly into the prompt, which sidesteps the retrieval contest altogether
How the same paid reach can condition the corpus a model actually reads, by amplifying third-party proof into the exact sources providers are now licensing directly
What to do with the rare, highly qualified traffic that does leak back from an AI-cited answer
Why an account graph is what turns all of this from a hunch into something you can target, prove, and improve
Today I am exploring an enterprise-grade workaround to the dark funnel problem. Not a set of challenger brand style gorilla tactics to get cited today and gone tomorrow, nor some reverse engineering style thinking to understand which accounts have seen you in an LLM - but a credible, and powerful way to influence the prompts that buying committees use to pre-=condition them to include your brand on their shortlist.
The dark funnel just moved somewhere you genuinely can’t see it
We’ve talked about the dark funnel for years, that stretch of research happening away from your owned properties, invisible to your CRM and your intent tooling.
What’s changed is that it’s now happening inside a genuinely closed conversation. No cookie. No pixel. No retargeting tag. No leaked query string to mine. A buyer opens a chat, forms an opinion about your category, and you get nothing back, not even the fact that the conversation happened.
The scale of it is the part worth taking note - the TLDR is that, pretty much, every B2B journey is now impacted by LLMs. This isn’t niche, you should assume every buying journey is impacted.
G2’s March 2026 survey of 1,076 B2B software buyers found 51% now start their research inside an AI chatbot rather than a search engine, up from 29% a year earlier. Forrester’s global numbers put overall usage higher still: 94% of B2B buyers used AI during their most recent purchase process, up five points on the year before [2]. These stats are pulling on different threads - G2 is saying the journey begins in AI more often than not, and Forester is saying that essentially all buying committees then use it at some subsequent point in the research journey afterwards.
The above usage stats don’t surprise me one bit - but this one does. Of those buyers, 69% ended up choosing a different vendor than the one they’d originally planned to, and a third bought from a vendor they’d never even heard of before that research began [1]. Research has long shown that the vendors that start the process at the front of the list often end the process with the DocuSign safely stowed in their inbox. Now it seems that buying committees are trusting the LLMm so much that it can, and the majority of time does, change their mind. So essentially the LLM is another member, perhaps the most influential one to boot, of the internal buying team.
That’s not a future-state slide. That’s the shortlist being rewritten, right now, inside the accounts on your target list, in the one part of the journey that has always been the hardest to reach and has just become instrumentation-proof into the bargain.
We have to get the LLM on-side, but fortunately we also know that any LLM is incredibly influenced by the prompt/s that go into it, and frankly they have a high agreement rate with the sentiment of the prompt.
I’m not saying any buying committee member would be this excitable in their prompt, but here is Perplexity being steered by the prompt
But this exaggerated point aside - how can we influence LLM’s and the incoming prompts with targeted approaches, which can sit alongside earning more citations for less steered prompts - lets dive in
Below is how I’d use paid reach to get a brand into buyers’ prompts and the sources AI models read.