Most companies think they are transforming go-to-market with AI.
Most of them are adopting it.
The difference is not how many tools they have bought. It is whether anything about how the organisation works has changed.
1. Adoption Is Not Transformation
Look at what usually counts as AI GTM.
A new copilot for reps. An AI SDR. AI-generated content. Search optimisation for AI answers. Automated workflows between systems.
All of it useful. Some of it very useful.
Now look at what stayed the same.
Same org chart. Same owners. Same handoffs.
Same KPIs.
The people do their jobs faster. The jobs
are the same jobs.
That’s adoption.
Adoption is not a failure. It is often the right first move. But it is easy to mistake for transformation, because it produces activity, dashboards and vendor case studies.
Transformation produces something else first. Discomfort.
2. The First Signal Is Organisational Discomfort
Real transformation rarely announces itself with a strategy deck. It shows up as a question nobody can answer.
Take AI search.
Buyers start asking AI assistants which vendors to consider. Organic traffic falls. Pipeline follows. The leadership team asks a reasonable question: who is responsible for whether we show up in AI answers, and for the pipeline that comes from them?
The honest answer is everyone and no one.
Product marketing owns positioning and the answers to buyer questions. Content produces the assets. The web team decides
whether AI crawlers can find any of it. Demand owns conversion. Customer marketing owns reviews and case studies. PR builds the third-party authority that AI systems cite. RevOps owns attribution, which doesn’t exist yet for this channel.
Seven functions touch the outcome. None of them owns it.
That is the first real signal of transformation: the old division of labour stops mapping to the outcome.
Nothing was wrong with any single function. Each was doing its job. The problem was the space between them, and until now
nobody had to look at it.
3. AI Didn’t Create the Fragmentation. It Made It Expensive.
The seams in a GTM organisation are not new.
Work has always crossed functions that don’t share an owner. What kept it working was people. Meetings. Handoffs. Judgement. Informal coordination. The heroic operator who knows who to call and
quietly fixes the gap before anyone notices.
Humans were the connective tissue, and they were good at it. Good enough that the fragmentation never showed up as a cost.
Then AI changes something faster than the organisation can coordinate.
The seams become visible. And because they now cost pipeline, they become expensive.
AI didn’t create the fragmentation. It made the fragmentation expensive.
To be precise about the claim: AI is not the only thing that can do this. A new CMO with a finance mandate, a new owner demanding accountability, a downturn: any shock exposes the GTM you actually
have. What is different about AI is that it is the first shock hitting every
company at once.
It comes in through two doors.
Through the buyer. AI changes how buyers discover, research
and shortlist. Nothing inside the company changed, but the outcome moved, and the organisation built for the old buyer journey no longer maps to the new one.
Through the work. AI changes how work gets done. Tasks get
automated, steps get skipped, outputs arrive faster than the next team can absorb them. The handoffs designed for human speed start to jam.
Which door it comes through depends on
where your buyer’s behaviour or your workflow hits your architecture first.
That is also where your first seam will show.
This is the same failure, one level up, as a problem I have written about in AI systems. There, agents reconstruct context that the architecture never kept. Here, people reconstruct coordination that
the organisation never designed. In both cases AI exposes the gap by moving faster than the humans who used to fill it.
4. Transformation Redesigns Around the Outcome
Once the seam is visible, there are two responses.
The first adds AI to each function and hopes the outcome improves. Content gets an AI writing tool. The web team gets an optimisation plugin. Demand gets an AI-search report. Everyone is busier.
Nobody owns the result.
The second redesigns the work around the outcome.
When companies actually do this, the changes tend to arrive in a recognisable order.
Ownership moves first. Someone is made accountable for the
outcome, usually whoever is closest to the buyer’s questions, and that person gets a pipeline number. In AI search that is often product marketing, which until now was rarely measured on pipeline at all.
Measurement changes before roles do. Impressions, click-through
and content downloads stop being the scoreboard. New questions replace them. Do we appear in the AI-generated shortlist? Are we described the way we want to be? Which buyer questions do we own? How much pipeline does this produce?
Then roles mutate. Few jobs disappear outright. The content manager becomes responsible for AI visibility. Product marketers each
report on how their product shows up. Marketing operations grows into someone who tracks AI referrals through to revenue. Some capabilities get built in-house, some get hired, and the build-or-hire choice is itself revealing.
Work moves between functions. Content writers move closer to the subject-matter experts. Demand stops producing content and starts aligning campaigns to what buyers ask AI.
Temporary governance appears. A weekly steering group, a
shared board, a scorecard, one accountable leader. That is coordination made formal, and it matters, but it is scaffolding, not the building.
Notice what doesn’t have to change: the org chart.
Plenty of real transformations leave the reporting lines exactly where they were. Functions keep their managers and take briefs from the new owner of the outcome. Sometimes the whole response is the
same team, with a new focus and a new goal, and no new headcount.
And plenty of fake transformations redraw the chart, rename a team “AI Ops” and change nothing else.
The unit of transformation isn’t the tool, or the org chart. It’s the operating model: who owns the outcome, how work moves between people, and what gets measured.
5. Exposure Travels
Fixing one seam reveals the next.
A company that wins back AI visibility discovers that visibility was never the goal. Showing up in the shortlist is table stakes. The new question is whether the buyers who arrive from AI convert, and that question doesn’t belong to marketing.
Those buyers arrive differently. They have already researched. They have already compared. They show up later in their journey, with sharper questions, and a sales process designed for buyers who
needed educating meets buyers who don’t.
So the exposed seam moves one handoff downstream: from visibility to conversion, then from marketing into sales.
AI exposes the seams in sequence, starting wherever the buyer’s behaviour changed first.
That is why most AI GTM stories today are marketing stories. It isn’t because marketing is further ahead. It’s because discovery is the first place buyer behaviour changed. Evaluation is next. Then the sales conversation itself, then renewal.
A company that treats its first fix as the finish line will meet the same problem again, one function along, and call it a new problem.
6. The Test: Could You Go Back?
All of this suggests a simple diagnostic.
If you switched AI off tomorrow, could you go back to how you worked in 2023?
If the answer is yes, you adopted. You would stop using the copilot, the SDR agent and the content generator, and the organisation underneath would carry on as before. Slower, perhaps. But the
same.
If the answer is no, you transformed. The things that would stop you going back look like this:
• These roles didn’t exist before, and the old ones they replaced are gone.
• This work moved to a different function.
• This human handoff no longer exists, and nobody is left who knows how to do it.
• This team now owns a different outcome, with a number attached.
• This metric didn’t exist, and the ones it replaced are no longer reported.
• This workflow runs without a person in the loop.
• This capacity was never hired, because it no longer requires headcount.
It is the same logic as the Removal Test for products, turned on the organisation. A product is embedded when taking it out would break something. An operating model has transformed when taking AI
out would break it.
The test also separates the permanent from the temporary. Run it on the changes described above. Would the new owner give
back the pipeline number? Probably not, so that change is permanent. Would the weekly steering group survive? Probably not, so that was scaffolding. Both were necessary. Only one of them is transformation.
7. Five Layers, and Where the Economics Belongs
Transformation goes deeper in layers:
Work → Roles → Architecture → Measurement → Economics.
AI changes the work. Changed work forces changes in roles. Changed roles, together with automation, force changes in how the organisation is put together: who owns what, and how work moves between people. A changed architecture needs new measurement. Eventually, the cumulative consequence shows up in the economics.
These are layers of depth, not steps in a sequence. A transformation can begin at any of them. When AI comes in through the buyer, ownership and measurement usually break first and roles follow
later. When it comes in through the work, roles move first. Wherever it starts, it isn’t finished until it reaches the economics.
Which is where the revenue-per-employee argument belongs, and it needs to be stated carefully.
The weak version: RPE went up, so we transformed.
It doesn’t hold. RPE also rises when a company simply cuts headcount, and a reduction in force raises it once without
changing how a single piece of work is done.
The strong version: AI GTM transformation should eventually change the relationship between human input and revenue output. Otherwise, what transformed?
The test that tells the two apart is one.
Question: did output hold or grow because the work changed, or because the people left? Cutting raises RPE once. Transformation keeps raising it, because capacity no longer scales with headcount.
Most companies I see haven’t reached that layer yet. The furthest along have it on the roadmap: AI agents running the reporting, workflows rebuilt to run without a handoff, capacity that no longer
needs a hire. That is the honest state of the market, and it’s the clearest measure of how early we are.
Where This Ends
The companies furthest ahead in AI GTM won’t necessarily have the most AI tools.
They’ll be the ones whose organisations no longer look as though AI never happened.
