Leaner Or Just Smaller

At the twenty five to a hundred million revenue mark, companies with high AI adoption are running go to market teams of about 45 people. Everyone else is running 65.

That is from ICONIQ's State of Go-to-Market report, dated March 2026, off the back of more than 150 B2B software companies. Across the revenue bands the gap runs somewhere between 23% and 31% smaller.

It is a real number and it is a big one. Twenty people is a lot of payroll.

The sentence I keep hearing

"We're running leaner with AI."

I've heard some version of that from four or five founders this year and I'll hear it again next month. And look, you're not alone in saying it. I know that doesn't help.

Because when I ask what actually changed, the answer is almost always the same two things. We made some redundancies. We bought a tool.

That is not running leaner. That is being smaller and hoping.

Those aren't the metrics I want to hear either. Headcount saved is a cost line, and anyone can cut a cost line, I could do it this afternoon. What's the revenue per rep now compared to twelve months ago? Has the win rate moved, or are you just running fewer deals? How much genuinely new pipeline got created last quarter, as opposed to recycled from the CRM? Because if the answer to all three is "about the same, but with fewer people", you haven't found leverage. You've found a smaller version of the same problem.

The number that ruins the easy story

Here is the bit that gets left out of every post quoting the ICONIQ figures.

Kyle Poyar's Growth Unhinged, published on 20 May, ran Sumble's job posting data across US digital natives. Go to market job posts were down 15% year on year in the first quarter. SDR and BDR posts specifically were down 21%. Customer support was down 37%.

So far, so obviously "AI is eating the entry level".

Except AI-native companies, the ones actually building this technology, more than doubled their SDR headcount over the same period. They accelerated go to market hiring by nearly 50% year on year.

Read that twice. The companies closest to the tools are hiring more salespeople, not fewer.

Now, does that mean AI isn't changing the work? No, obviously it is. Poyar's data also has go to market engineering headcount doubling, with 400-plus of those roles now sitting at US digital natives, which is a job title that barely existed two years ago. And three in five open go to market roles are account executives or solution engineers, so the shape of the team has genuinely shifted towards people who can run a deal rather than people who can send a sequence.

But "the fastest growing companies in the category are doubling their SDR teams" and "AI replaces SDRs" cannot both be true. One of them is a story and one of them is a job posting.

What actually produces it

I want to be careful here, because I'm not arguing the leanness is fake. ICONIQ's 45 against 65 is real and it is measured.

The question is what produced it.

At the companies where it works, the tooling changed how the work gets done before the headcount changed. Research that took a rep forty minutes takes four. The list gets built from actual signals rather than a filter on Sales Navigator. The follow-up doesn't fall over because someone was on holiday. Then, some months later, you find you don't need to backfill the two people who left, and your team is 30% smaller without anyone deciding it should be.

At the companies where it doesn't work, the order is reversed. Cut first, buy a tool, tell the board it's an AI efficiency programme, and hand the same broken process to fewer people with a licence each.

There's no silver bullet in this, otherwise we'd all be doing it. But the order matters enormously and almost nobody talks about it.

Both failure modes are the same failure

One of my clients hired a proper, full-time BDR last year. Spent £110,000 on him. Got zero leads.

That was not an AI problem. There was no AI anywhere near it. It was a hiring decision made in place of a system, and it failed for exactly the same reason the redundancy-plus-tool version fails: nobody had decided who they were selling to, what the message was, or what a good week looked like before they went and bought capacity.

Hire a body without a system and you get nothing. Buy a tool without a system and you get nothing, faster and cheaper. The tool is not the variable.

And the whole outbound thing is so fried at the moment that neither approach survives contact with a real inbox on its own.

One number to treat carefully

You will see a stat going round that 36% of B2B companies cut their sales development teams in 2025. That comes from a SaaStr and Emergence Capital survey of 560-plus B2B software companies, and I've only seen it quoted second hand, in a piece syndicated in August, rather than in the original survey. I think it's probably about right. I wouldn't build an argument on it.

I'd also say: if you cut your SDR team in 2025 and your pipeline is now thin, the honest read is not that AI failed you. It is that the SDR team was doing something, and nothing replaced it.

The lesson

Smaller is not the same as leaner. Leaner means the same output from fewer people, and you can only get there by changing the work first.

If you've cut heads and bought a tool and nothing else about how your team sells has changed, you haven't adopted AI. You've just got a smaller sales team and a subscription.