Why AI Didn't Fix Your Marketing

AI is making you slower. You just can't feel it.

I know how that sounds. Every headline says the opposite. But this isn't an opinion, it's been measured, and the same blind spot that makes AI feel fast is the exact reason your marketing still isn't working.

Stay with me, because this one changes what you do next.

You started a coaching business and found out it came with a second job nobody warned you about: marketing. So when AI showed up promising to finally take that job off your plate, you did the responsible thing. You dove in. ChatGPT. Claude. The saved prompts, the tutorials, a custom GPT or two. Exactly what a smart, capable person does when the ground shifts under them.

And somewhere in there, quietly, AI handed you a third job. Not coaching. Not marketing. AI Engineer. Prompt-writer, tool-tester, output-checker, the person who has to learn a new platform every week just to keep up.

Three jobs. You only ever wanted one. And after all of it, your bank account looks about the same.

Let me show you what actually happened because most people are missing this.

Using AI feels like progress. You type a request, and in four seconds there's a full draft on the screen. That feeling of speed is real. It's also misleading.

A research group called METR ran a careful experiment on this. They took experienced developers and had them do real work, some of it with AI tools, some without. Before they started, the developers predicted AI would make them about 24% faster. The result: they were actually 19% slower with AI. And the part you need to sit with, even after finishing the work slower, they still believed the AI had sped them up by 20%.

Read that again.

That's not a knock on them. It's how the tool works on all of us. The draft appears fast, so your brain logs "fast." The hours you then spend fixing the generic parts, re-explaining your context, rewriting it to sound like you, cutting the stuff that's confidently wrong, those hours are spread out in little pieces, so you never add them up. The speed is visible. The cost is hidden.

Upwork's research institute found the same thing in the real workplace. 96% of executives expected AI to make their people more productive. Then they asked the people. 77% said AI had actually added to their workload. The tool didn't remove work. It changed the kind of work, and quietly gave them more of it.

That's the third job, measured.

You can feel busy and productive and still not move the only number that matters. You're posting more than you ever have. You're producing. And the calendar is still quiet.

If that's you, I want you to hear this clearly: it is not a you problem. The biggest companies in the world, with real budgets and full data teams, got the exact same result.

This year MIT published a study with a title that says it all: The GenAI Divide. They looked at enterprise AI across hundreds of deployments. The finding: after an estimated $30 to $40 billion in spending, about 95% of companies saw no measurable impact on their bottom line. Not a small return. No measurable return. Only around 5% of those projects produced real results.

It's not just MIT. McKinsey surveyed nearly 2,000 organizations across the world. Almost everyone is using AI somewhere. Only about 6% have turned it into real financial value. S&P Global found that the share of companies abandoning most of their AI projects jumped to 42% in a single year, up from 17%. RAND found that more than 80% of AI projects fail, roughly twice the failure rate of normal technology projects.

Billions of dollars. Smartest engineers money can hire. And the needle didn't move.

So if a Fortune 500 with a hundred-person team couldn't make generic AI produce results on its own, the problem was never that you didn't find the right prompt.

The tool isn't broken. It's just being asked to do something it can't do alone.

Generic AI is a brilliant impressionist. Ask it for a marketing plan and it will hand you a polished, professional, confident-sounding answer in seconds. The problem is where that answer comes from. It's starting from a blank page and pulling the most common, most average response from everything it has ever read. Harvard Business School tested this directly: they asked the top models for strategy advice across dozens of different companies, and the models gave back the same fashionable answers regardless of the company. Adding more detail barely changed it.

That's the catch. AI doesn't know your client. It doesn't know your offer, your story, your market, or why people actually buy from you. It gives everyone the same answer, dressed up to look custom.

And when you can't tell the difference, it gets dangerous. The same Harvard team, working with Boston Consulting Group, ran an experiment with 758 consultants. On tasks that sat inside what the AI is good at, the consultants using it did great: more done, faster, higher quality. But on a harder task, one that sat just outside the AI's real ability, the consultants using AI were 19% less likely to get the right answer than the ones working without it.

Think about why. The AI produced an answer that looked just as polished as always. Confident. Well-written. Wrong. And because it looked right, the experts trusted it and switched off their own judgment. The tool didn't lift them. It misled them, with a smile.

That is the whole thing in one sentence.

This isn't me being the lone skeptic at the AI party.

Ilya Sutskever helped build the technology behind ChatGPT. He co-founded OpenAI. In late 2025 he said the easy era is over. His words: the age of just building bigger models is done, and we're "back in the age of research," the hard, unsolved part. The people closest to this are openly telling you the magic-button version isn't coming.

And look at what the winners actually do. BCG studied the companies that are getting value from AI. Those leaders spend about 70% of their effort on people and process, and only about 10% on the AI itself. Let that land. Even the success stories will tell you the model is the smallest part of the job. The expertise around it is the job.

AI is a multiplier. That's the honest way to think about it.

Point it at expertise you already have, and it makes you faster and sharper. Point it at expertise you don't have, and it makes you faster at producing something that misses. The multiplier doesn't care which one it's amplifying. That's why the same tool turns a skilled marketer into a powerhouse and turns a brilliant coach into someone with a third job and a quiet calendar.

So the missing piece was never a better prompt, a new tool, or one more course on AI.

You were never supposed to become that expert on top of everything else. That was the wrong ask from the start.

So the real question was never "which AI tool should I use." It's "who is directing it."

Next week I'm going to show you the answer, not tell you. I'll walk you through what AI looks like when it's pointed at your business by people who already have the marketing expertise built in. An AI team that knows coaching, knows client acquisition, and then learns you, and builds the whole client system in front of you. It's the most fun issue in this series, and it's the payoff for everything we've covered.

For now, the first step costs nothing and takes a few minutes. Before you direct anything, you need to see clearly where your marketing actually stands right now. That's what the free Marketing Score does. It gives you an honest read on your profile, your content, your offer, and your messaging, and shows you the specific gap that's quietly costing you clients.

It takes a few minutes and costs nothing — here's where to start:

[ GET MY FREE MARKETING SCORE → ]

[ JOIN US LIVE to LEARN HOW TO INSTALL YOUR OWN A.I. MARKETING TEAM → ]

See where you stand this week. Next week, I'll show you the team that closes the gap.

Norbert Orlewicz
The Profitable Coach