Let me start with the budget reality, because everything else follows from it.
Gartner's 2026 CMO Spend Survey, which went out to just over 400 marketing leaders across North America, the UK and Europe, found marketing budgets sitting at 7.8% of company revenue. In 2025 it was 7.7%. That is a flat line. Meanwhile CMOs are putting an average of 15.3% of that budget into AI, and 70% say becoming an AI leader is critical this year. Only about 30% describe their organisation as actually ready to scale it. More than half say they do not have the budget to execute their strategy.
So: same money, new mandate, no readiness. Every CMO I speak to is living in that gap right now.
You cannot fund an AI transformation out of new money that is not coming. It has to come out of what you already spend, and in most B2B companies the biggest line is people. That is the real reason team structures are changing, and I would rather say it plainly than dress it up.
What the org chart looks like now
The numbers are consistent across sources. LinkedIn workforce data shows marketing job postings growing roughly 6% while total marketing output grew around 24% over the same period. Benchmarking work puts restructured teams at 15% to 22% fewer people than historical medians for their revenue band, producing more.
Median headcount by size, from 2026 benchmarks: companies between £1m and £10m in revenue run about three marketers. Between £10m and £50m, around eleven. Past £50m it jumps sharply.
Three people at £10m. I want you to sit with that, because I speak to founders every month who think they need six.
The role that disappeared is the one nobody wants to name: the mid-level single-channel specialist. The person who competently executed content, or email, or reporting, and nothing else. AI absorbed that layer first and most completely. What is left is barbell-shaped: senior people who decide, and sharp generalists who orchestrate systems.
Tools and platforms now run at 14% to 18% of fully loaded cost per marketer, up from around 9% in 2022. Money moved from salaries to systems. That is not a prediction, it already happened.
Where the marketing team went
LinkedIn · 2026 benchmarksThe question I now ask first
When I go into a company, I no longer start with 'what are we doing'. I start with a version of what analysts call a capabilities audit rather than a headcount audit: what does this team actually produce, and which of those outputs need a human brain?
You end up with three buckets. Volume work: drafting, resizing, list building, enrichment, transcription, reporting, first-pass research, translation. This belongs to systems now, with a human editing the output. If a person on your team spends most of their week here, you have a design problem, not a performance problem.
Judgement work: positioning. Pricing input. What to say no to. Which markets, which segments, which bets. Relationships with press, analysts, partners, customers. Anything where being wrong is expensive and being right requires context a model does not have.
The messy middle: campaign design, creative direction, sales alignment, event strategy. AI assists heavily, humans still own it.
Design the team around the second and third buckets. Buy or build systems for the first. That is genuinely the whole method, and it is much harder to do politically than technically, because the first bucket is where several people's job descriptions currently live.
The honest numbers on AI outbound
This is where I want to be careful, because there is more nonsense written about AI SDRs than about anything else in our industry.
The adoption is real. Around 41% of enterprise B2B teams had at least one AI SDR in production in the first quarter of 2026, up from a tiny base two years earlier. And some of the gains are real too. Benchmarks from Apollo and ZoomInfo show per-rep monthly outbound volume rising from roughly 1,150 to 7,400. The Bridge Group found cost per qualified opportunity falling from about $487 in human-only teams to $224 in hybrid teams. Ramp time for a new AI seat is measured in weeks rather than the four to five months a human SDR takes to get productive.
Now the other half, which vendors are less keen to lead with. Raw reply rates in the same benchmarks fell from 4.7% to 2.9%. Deliverability research from Smartlead and Instantly found domains running AI outbound at production volume losing sender reputation sharply within ninety days, and that domain reputation collapse now caps a large share of attempted deployments inside the first three months. Microsoft 365 inboxes are the strictest filter.
The mechanism is simple and worth understanding. One model, one system prompt, one set of value propositions produces output that converges on recognisable patterns, even with personalisation tokens. Filters cluster it. And software ramps instantly while sender reputation only ever builds slowly. As one deliverability engineer put it recently, agents do not break the physics, they just compress the timeline from quarters to weeks.
In head-to-head comparisons, human SDRs still generated substantially more revenue than AI-only equivalents and got far better meeting show rates. The model that works, and that roughly 45% of teams have landed on, is hybrid: machines do the research and the admin that eats most of an SDR's week, humans do the conversations.
My rule with clients is blunt. If your plan is more emails, you have not understood the tool. If your plan is better emails to better-researched people at the same or lower volume, now we are talking. Cost per opportunity fell by about half in the good implementations. It did not fall by 90%, and anyone promising you that is selling something.
Human-only vs. hybrid AI outbound teams
Apollo · ZoomInfo · The Bridge GroupWhat I would actually build this year
If you gave me a B2B company between £5m and £30m in revenue and a flat budget, here is the shape.
One senior owner of the number. Full-time if you have repeatable revenue and a team to manage. Fractional if you are still deciding what marketing should be. I have done this both ways and the failure mode is always the same: hiring a full-time leader to figure out something that was actually a twelve-week project.
Two or three sharp executors. Generalists who are comfortable running systems. Hire for judgement and curiosity over channel CVs.
A systems layer you own. Research, content production, outbound, reporting, CRM, all wired together and running in your accounts on your data. Not a black box, not a dependency on an agency, not six point tools that do not talk to each other. Forrester expects a majority of revenue leaders to consolidate today's six to eight tool stacks into single orchestrated systems within eighteen months, so build with that in mind.
Bought capacity for spikes. Design, video, paid media, events. Buy it, do not hire it, until the volume is permanent.
And a governance line. This one is new and it matters commercially: Forrester's 2026 survey found 68% of enterprise buyers now factor a vendor's AI governance into technology purchasing decisions. How you use AI is becoming a thing your customers ask about in procurement. Write the policy before someone asks for it.
Part three is about the thing all of this leverage cannot buy, which turns out to be the expensive part.
Frequently asked questions
How big should a B2B marketing team be in 2026?
Median benchmarks put companies at £1m to £10m revenue on about three marketers and £10m to £50m on around eleven, with restructured teams running 15% to 22% below historical medians while producing more.
Do AI SDRs work?
In hybrid form, yes: cost per qualified opportunity roughly halves and ramp time collapses. AI-only outbound at high volume tends to damage deliverability within ninety days and produces lower reply rates.
Where should marketing budget go if it is flat?
Out of volume execution and into systems and senior judgement. Tools now account for 14% to 18% of fully loaded cost per marketer, up from about 9% in 2022, and that shift funds itself.
Which marketing roles has AI changed most?
Mid-level single-channel execution roles. Strategy, positioning, relationships and creative direction are unchanged or more valuable.