What six months taught us about how fast the market moves now, and the four shifts we didn’t see coming.
In December we published five predictions for how B2B go-to-market would change in 2026. We expected to revisit them in a year. We’re revisiting them in six months, because all five already happened. The forecast wasn’t wrong. It was slow.
We said:
ICPs would get narrower and evidence-led. They did, and the stakes rose. Teams that let an AI tool pick their accounts now inherit the same targeting as every competitor on that tool, so deciding which segments to win, and why now, became the scarce work. “Financial services, 500+ employees” is an embarrassment. The ICP that wins lives in your systems, not on a slide.
AI would orchestrate GTM, not just write content. It does. Anthropic runs its own go-to-market with AI as the orchestration layer, and Snowflake runs agents across a 700-person team that cut cost per opportunity by 30% in six months.[1] Agents qualify, route, and sequence all day.
Martech stacks would shrink and consolidate. They are, in public. Swan killed 27 of its 30 agents and still scaled, and Snowflake now certifies every shared skill centrally before anyone runs it. The flex stopped being how many tools you own and became how few you trust.
Content would be built for GenAI discovery. This one we undershot. In early 2026, 68% of US searches ended without a click,[2] and most of what AI cites was never your site. It didn’t stay a content tactic. It became the funnel. More on that below.
GTM would run as an experiment system, not a plan. It does. The strongest teams now run a portfolio of proven and experimental motions and retire what stops earning its place. The catch we didn’t flag: the scoreboard for measuring it broke. More on that below too.
Five for five, in two quarters. Hold that thought.
A year-long forecast that resolves in six months isn’t a lucky guess. It’s a clock change. One of the sharpest operators we read put a number on it: the resting heart rate of the market went from 60 to 120. The cycle that used to take a year now takes a quarter.
The play that got you here will not get you there. Which is why the more useful half of this post isn’t the five we called. It’s the four we didn’t.
We predicted AI in discovery. We underestimated what that meant.
The recommendation became a purchase trigger, not a research aid. Scrunch’s clickstream study, which joins a panel to people’s actual ChatGPT, Claude, and Gemini conversations, found that an AI recommendation to someone who had never heard of a brand more than doubled their chance of searching for it within a week, from 2.9% to 7.2%. A recommendation moved behavior two to three times harder than a neutral mention.[3]
And it happens off your own domain, before a salesperson exists. Roughly 90% of what AI cites for non-branded questions comes from third-party sources, not your website.[4] First touch and the shortlist now form somewhere you don’t control, in a conversation you never see.
Then the part nobody forecast in December: a paid layer is forming on top of the answer. OpenAI crossed $100M in annualized ad revenue within weeks of opening ChatGPT ads.[5] The first impression, the shortlist, and now the ad all happen before a click.
Call it the answer-layer market. The website’s new job is to be the source the machine quotes, not the place the buyer makes up their mind.
When execution costs close to nothing, the scarce thing isn’t output. It’s the call of what to say, to whom, and what not to automate.
The companies that understand this are paying for it. AI-native firms are hiring senior humans to own narrative, including a Head of Copy role posted at $320,000 to $400,000 and brand leadership roles framed as creation, not stewardship.[6] These are the same companies that automated everything else first. They are not paying that for output. They are paying for judgment.
Expertise is the only thing in GTM that isn’t a commodity now. AI gets a team to average. Average is exactly what the model already produces for free.
Speed is solved. Agents produce all day. What doesn’t compound is the foundation they read from.
The teams pulling ahead have what one operator calls a context layer: a persistent, structured spine of ICP, positioning, messaging, and brand that every output draws on and writes back to.[7] Without it, every new asset starts from scratch, and none of it makes the next one sharper.
The test is simple and most rooms fail it. Point at one artifact that is making every other artifact better. Silence.
AI scales whatever motion you give it. Including the wrong one. The constraint was never the agents. It’s the thing worth pointing them at.
Here’s the catch under prediction five. GTM did become an experiment engine. But the scoreboard broke.
The channel now driving qualified discovery resists clean attribution. AI recommendations shift by as much as 80% when a single setting, web search, is switched on, and the same prompt won’t reproduce the same answer twice.[8] You cannot optimize once for every engine, and you cannot prove the result the way a board wants it proven.
Meanwhile the proof gap is widening. Jasper’s 2026 survey of 1,400 marketers found 91% now use AI, but only 41% can show a return, down from 49% a year earlier.[9] The distance between works and provably works is the story of the next two quarters.
The market’s answer is the oldest one there is: proof in public. The real number, with the failure left in.
Not everyone agrees this is new. Amos Bar-Joseph, who runs Swan, calls the answer-layer shift rebranded SEO: the surface changed, the decision didn’t. He has a point worth holding. Much of this is old marketing logic, reputation, trust, being known, wearing new mechanics. But the mechanics moved the first touch off your domain and put a meter on it. New logic or not, that’s where the buyer now is.
So here is the takeaway worth keeping. The cycle compressed, and it is not slowing down. The teams that win the next six months won’t be the ones with the most agents. They’ll be the ones with the foundation worth pointing the agents at.
That’s the work. At Forabilis, we run your GTM with you: the strategy that holds, the marketing engine that runs, the growth that follows. Whether you build that foundation with us or without us, it can’t wait for 2027.
Link copied!
