AI Raised the Floor and Lowered the Ceiling
Olly Jones on why AI made mediocrity accessible — and the metric that catches the damage before your reputation does.
Olly Jones has a line that captures the last year of AI in marketing better than most reports: AI raised the floor and lowered the ceiling.
The floor came up. Teams that used to ship bad work now ship competent work. But the ceiling came down with it, because the teams that used to make something remarkable saw the shortcut to competence and took it. So now everyone serves the same vanilla, and buyers can’t tell anyone apart.
Olly watches this as a commercial operator, not a marketer. He spent 10+ years at American Express running $3B+ global accounts, founded and exited a tech services company, and now works as a fractional CRO across 15+ startups a year — while running the Forward Deployed Marketer division at Myosin. His lens is revenue, which means he tracks the number one AI marketing content skip: what all that production actually costs to turn into a customer.
We sent him our questions. His answers are below, lightly edited as always.
You spent a decade at American Express running $3B+ global accounts, then built and exited a company. What did closing enterprise deals at that scale teach you that most AI-era marketing advice gets wrong?
At AmEx, I learned fast that you could show a VP how to save their company millions of dollars, and they still might not move on it — because the decision didn’t help their career. Maybe it made another department look good. Maybe it created political risk. Maybe it didn’t align with what their boss cared about that quarter.
You have to understand what that individual person needs to succeed, not just what the company needs.
Most AI-era marketing advice treats buyers like rational economic actors. Here’s the ROI. Here are the cost savings. Here’s the efficiency gain. But enterprise deals don’t close on spreadsheets. They close on trust, internal politics, and whether the buyer can take credit for the win.
We’ve also lost the face-to-face context post-COVID. Before, I’d meet stakeholders in person every week. You’d pick up on body language, side conversations, who deferred to whom in the room. You’d understand what someone really wanted versus what they said on the call.
Now everything’s remote, and you lose that context. Standard AI tools make it worse — they optimize for volume and efficiency, not for reading the room or understanding the human on the other side.
The companies winning enterprise deals today are the ones that still prioritize the individual buyer’s agenda, not just the company’s problem.
You’ve said most AI gives teams 70th-percentile output, and your work is helping them hold 95th-percentile thinking at scale. What’s the actual difference in practice — and where does a smart team accept the 70th percentile without realizing it?
AI can write copy. It can summarize data. It can generate outreach sequences. But it struggles with why this matters to this specific person.
70th percentile positioning sounds fine on the surface. It’s grammatically correct. It uses the right keywords. It checks the boxes. But it’s generic. It tries to appeal to everyone, so it resonates with no one.
95th percentile positioning is sharp. It’s specific. It knows exactly who it’s for and what problem it’s solving. It doesn’t hedge. It doesn’t try to be all things to all people.
My favorite analogy is ice cream. A cheap knockoff Cornetto, three scoops of vanilla in a cup, and an extravagant sundae are all technically ice cream. But they’re not the same.
AI raised the floor: people who used to sell crappy fake Cornettos can now produce decent scoops of vanilla. That’s progress. But AI also lowered the ceiling: the companies that used to make beautiful, differentiated, memorable sundaes started thinking, “Why put in all that effort when we can serve vanilla scoops like everyone else? It’s faster, cheaper, and people still buy it.”
Now everyone’s selling vanilla ice cream in a cup. Things got dull. Hard to differentiate. And the buyers are overwhelmed by sameness.
The trap: AI makes it easy to produce “good enough” content at scale. But good enough isn’t remarkable. It doesn’t stand out. It doesn’t get shared. It doesn’t close deals.
The opportunity: the companies still making sundaes — the ones investing in positioning, storytelling, and depth — are the ones people remember. AI didn’t kill differentiation. It just made mediocrity more accessible.
You help companies use AI without falling into the traps everyone else does. Name the three you see most often from teams that look sophisticated from the outside but aren’t actually working.
Trap 1: Bolting AI onto broken processes. Most companies add AI to systems that already don’t work. They automate bad workflows and make them faster. Your sales process is unclear, your ICP is fuzzy, your messaging doesn’t resonate — so you add AI to scale outreach. Now you’re sending 10x more emails that still don’t convert and are potentially degrading your brand. AI didn’t fix the problem. It amplified it.
Trap 2: Optimizing for volume over quality. AI makes it easy to produce more — more content, more outreach, more campaigns. So teams do. But nobody’s asking whether they should be producing this in the first place. I see companies publishing 20 blog posts a month that nobody reads, sending 5,000 outreach emails with 0.2% reply rates, running campaigns that generate leads but not revenue. Volume feels like progress. But if the output doesn’t move the business forward, it’s just noise.
Trap 3: Focusing on how content is produced, not how it’s received. This is the one I see most right now. Teams obsess over tools, workflows, and prompts. How can we produce this faster? How can we automate it? But they’re not asking: if I were on the receiving end of this, would I care? There are hundreds of tools built to help people produce AI content. Almost none focused on how people receive it and use it to make decisions. If your AI-generated email looks like every other AI-generated email, the recipient deletes it. Before you hit send, ask: if I received this, would I actually read it? Would it change my behavior? If the answer is no, don’t send it — it doesn’t matter how efficiently you produced it.
A lot of AI marketing content chases output volume — more content, more emails, more outreach. As a commercial operator, where does that lead to the worst outcome for a CMO, and what metric actually captures the damage?
The worst commercial outcome: burning through your contact list without understanding cost per acquisition. I’ve seen companies send hundreds of thousands of outreach emails without ever looking at their CAC. They’re optimizing for activity — leads generated, emails sent, content published — instead of outcomes. And by the time they realize the economics don’t work, they’ve burned their reputation with their target market.
The metric that catches the damage is cost per acquisition. Most companies track top-of-funnel metrics: open rates, click rates, leads generated. But they don’t track how much it actually costs to acquire a customer. If your CAC is climbing while your volume increases, something’s broken — you’re reaching more people, but the quality is dropping, your messaging isn’t resonating, or you’re attracting the wrong audience.
The other metric I watch is reply sentiment. Not just reply rate, but what people actually say when they reply. “Not interested” or “unsubscribe” at scale means your messaging is off. “This doesn’t apply to us” means your targeting is wrong.
There’s also an opportunity to improve at allocating tokens to specific initiatives and measuring ROI from that allocation. Right now, most companies have AI running in the background, racking up costs without understanding what they’re getting for it. It’s like leaving the lights on in every room of the house. Start tracking which AI workflows generate revenue and which generate noise, and allocate accordingly.
Pricing is one of the hardest things for B2B founders past PMF, and AI is making it messier—value-based, usage-based, and AI feature add-ons. What pricing mistake do you see most often right now, and what’s a simple test you’d run before changing a model?
The biggest mistake: trying to extrapolate SMB pricing into an enterprise model. I see this constantly. A company has a $150/month SaaS product for individuals or small businesses. It works. Then an enterprise buyer shows interest, and instead of rethinking the model entirely, they just scale it up. “$150/month × 100 seats = $15K/month enterprise plan.”
But enterprise buyers aren’t buying seats anymore. They’re buying outcomes, services, integrations, workshops, and support. AI brings too many personalization opportunities for seat-based pricing to hold. Your product might be part of the solution, but it’s not the whole solution. If you’re still charging per seat when the enterprise buyer is thinking about business outcomes, you’re creating confusion, not clarity.
The simple test I run before changing a pricing model: put yourself in the buyer’s shoes and imagine explaining the purchase to their boss. If the buyer struggles to explain the value clearly, or the justification feels weak, your pricing model is wrong. Good pricing makes the buyer’s internal pitch easy and makes them look like a star. Bad pricing makes them do mental gymnastics to justify it.
“Forward-deployed AI marketing” is a sharp phrase. What does it actually look like in execution for a B2B company past PMF — the team, the tooling, and week one?
Forward-deployed means we’re embedded, not advising from the outside. A typical setup is two FDM operators — we find that works much better than one or three-plus builders — plus an account manager so the operators focus on building, not project admin, and internal stakeholders on the client side: a CMO or growth lead and a sales lead. Weekly sync, daily async.
We don’t bring a pre-set stack. We work with what they have and add only what’s missing. Common tools in the mix: Clay for data enrichment and outbound; Claude or GPT for content and research synthesis; Relay or Make for workflow automation; and their existing CRM—Salesforce, HubSpot, Pipedrive, whatever they use. The philosophy is to build systems they own, using tools they can run after we leave. We’re not creating a dependency. We’re building infrastructure.
Week one starts with a tech stack audit. Most companies have AI-native tools like Clay or traditional SaaS with new AI capabilities like Salesforce that are massively underutilized. We map what they have, identify what’s being ignored, and almost always find quick wins — honestly, often around meeting transcripts, internal and external, that nobody’s reviewing. There’s incredible signal in those: customer pain points, competitive intel, positioning feedback, objection patterns. Most teams record the meeting, file it, and never look at it again.
Common week-one deliverables: tech stack rationalization (what you have, what you’re using, what’s wasted), a data source audit (especially underused sources like transcripts, CRM notes, support tickets), and quick wins we can surface in days, not months.
You’ve said traditional agencies put their A-team on the pitch and juniors on the work. With AI layered on top, what should a senior marketer evaluate differently when hiring an agency in 2026 — and what question are they not asking?
The question they’re not asking: are you using AI as leverage or as a replacement? Most agencies use AI to replace junior labor — automating the work that used to go to entry-level marketers: content writing, social scheduling, basic research. That’s fine if you’re optimizing for cost. It’s terrible if you’re optimizing for quality. AI produces 70th-percentile output by default. If you use it to replace humans without adding a judgment layer, you get average work at scale.
What senior marketers should evaluate in 2026: Who’s actually doing the work — not just who’s on the team, but who’s reviewing the AI output, making the strategic calls, ensuring quality doesn’t collapse? What’s their AI operating model — using AI to multiply senior capacity (good), or to eliminate junior capacity without replacing it with senior judgment (bad)? What’s their quality control process — how do they make sure AI content doesn’t sound like everyone else’s?
The other thing CMOs aren’t asking enough: how do I get ahead, not just catch up? I hear too many CMOs asking how to get up to speed, how to catch up. That’s the wrong question. Two years ago the same CMOs were talking about how they were getting ahead. Now there’s a perception — mostly driven by LinkedIn hype and demo videos — that everyone else is further along. And perception becomes reality.
You’re a fractional CRO across roughly 15 startups a year. When does fractional actually beat a full-time hire — and when have you talked a founder out of bringing you on?
Fractional beats full-time when you need to build the foundation or reset, not run it forever. Pricing strategy, ICP definition, sales process design, GTM infrastructure — that’s a 90-day sprint, not a full-time role. It also wins when you don’t yet have enough commercial complexity to justify full-time: under about £2M ARR, with a sales process still being figured out, a full-time CRO is overkill. And when you need senior strategic horsepower without the overhead — a full-time CRO costs £150K+ in salary, plus equity and benefits, whereas a fractional engagement delivers the same strategic output in a focused sprint for a fraction of that.
I talk founders out of hiring me fairly often, and I think that’s a good sign. Sometimes founders assume that doubling my time gets output twice as fast. But there’s a point of diminishing returns. When you increase a fractional person’s capacity, they start joining more internal meetings and get pulled into keeping-the-lights-on work instead of high-leverage work. I’ve had clients move me from 10 to 20 hours a week, and honestly, the output didn’t double. I just spent more time in meetings that didn’t make me more productive.
When a founder just needs execution, not strategy, I used to refer them to a marketing operator or an agency. Now I’m finding AI systems can sometimes fill that gap alongside me — second-brain tools, workflow automation.
You work across Dubai, Europe, and New York, advising on crypto, AI tools, and B2B SaaS. What’s a commercial shift you’re seeing that hasn’t hit the LinkedIn discourse yet — and what’s the implication for senior marketers paying attention?
The shift: using AI to go deeper, not just to produce more. There’s a lot of talk about using AI to produce more content, do more outreach, and run more campaigns. Not enough about using AI to think more deeply about why you’re producing it in the first place.
The best companies are using AI inwardly, not just outwardly. They use it to understand their own story better — why does this product exist? What problem are we actually solving? To surface insights from customer conversations they’ve already had. To go deeper on positioning and messaging, not just write more copy, but understand the why behind it. To think critically about commercial strategy.
The implication for senior marketers: stop using AI as a production tool. Start using it as a thinking tool. Instead of “how can I use AI to write 10 blog posts this week?” ask “how can I use AI to understand what our best customers actually value, so we can position more effectively?” Instead of “How can I automate more outreach?” ask “How can I analyze the last 100 sales calls and find the objections we’re not addressing?”
The companies that win with AI won’t be the ones producing the most content. They’ll be the ones using AI to think more clearly about what to produce and why it matters.
Olly Jones is a fractional CRO and the founder of GTG Studio, where he builds revenue engines for B2B startups across fintech, SaaS, crypto, and AI. He spent 10+ years at American Express managing $3B+ global accounts, founded and exited the tech services company Kunai, and advises 15+ startups a year across Dubai, Europe, and New York. As Senior Operator and Partner at Myosin — a certified Claude Partner — he leads the Forward Deployed Marketer division, embedding AI-native marketing specialists inside B2B companies on 90-day engagements.
This wasn’t a paid post. We interviewed Olly because he had something worth saying. We’re always looking for collaborations, case studies, and sharp marketers with insights to share. If that’s you — a leader with a good story — just hit reply. We’ll sort out the rest together.
Share with a colleague. Earn paid membership months.