AI Saved You 10 Hours a Week on Marketing. You Spent 15 More.

Spotify's top engineers haven't written a single line of code since December 2025.

They didn't get fired. They got promoted to directing an AI system called Honk, built on Anthropic's Claude Code. An engineer on their morning commute can tell the AI to fix a bug or build a feature via Slack, then review the finished code before arriving at the office. Spotify shipped over 50 new features in 2025 this way.

The story sounds like a productivity dream. AI handles the execution. Humans handle the thinking. Everyone goes home early.

Except that's not what happened elsewhere.

A Harvard Business Review study published the same week told the other side. Researchers from UC Berkeley spent eight months embedded at a US tech company, tracking 200 employees who adopted AI tools. Their finding: 83% of workers said AI increased their workload. Not decreased. Increased.

The tools didn't free people up. They filled people up. Workers took on tasks they'd previously outsourced. They worked at a faster pace, across a broader scope, for more hours in the day. Nobody asked them to. The AI just made it possible, so they did.

If you run marketing for a small business and you adopted AI tools in the last twelve months, this should sound familiar.

You Didn't Save Time. You Absorbed More Work.

Here's what AI marketing adoption actually looks like for most SMEs.

Before AI, you wrote two blog posts a month, ran three ad variations, and sent one email a week. It took most of your available marketing time.

After AI, you write eight blog posts a month, run fifteen ad variations, send three emails a week, and post across four social channels daily. It takes... most of your available marketing time. Plus weekends.

The work expanded to fill the capacity. This isn't a failure of willpower. It's a documented psychological pattern.

The UC Berkeley researchers called it task expansion: when AI removes friction from a task, people don't stop. They absorb adjacent work that was previously too time-consuming to justify. The product manager starts writing code because AI makes it possible. The business owner starts producing daily social content because AI makes it quick. The one-person marketing team starts running campaigns across five channels because AI makes it feasible.

None of these people asked for more work. The tool just made more work feel achievable. The HBR researchers found this pattern led to 12-hour workdays, cognitive fatigue, and weakened decision-making. The productivity surge gave way to burnout.

The numbers across marketing are stark. 94% of marketers plan to use AI for content creation in 2026. Content production volume has roughly doubled for teams using AI tools. And yet Socialinsider's analysis of 70 million posts found overall engagement fell approximately 24% year-on-year. Instagram's median engagement rate dropped from 2.94% in January 2024 to 0.48% by 2025.

More content. Less response. More hours. Same results.

More Marketing Activity Doesn't Mean More Results. The Science Says It Makes Things Worse.

This isn't just a content glut problem. It's a strategy problem, and the marketing science explains exactly why.

Les Binet and Peter Field analysed 996 campaigns from the IPA Effectiveness Awards spanning three decades. Their central finding: the optimal budget allocation for long-term growth is roughly 60% brand building, 40% sales activation.

What happens when you flip that ratio, spending most of your effort on short-term activation? Strong initial results that erode year after year. Customer acquisition costs rise. Margins collapse. You need to spend more just to maintain the same volume of leads. Binet's data shows brands that invest nearly 100% in performance at the expense of brand building push acquisition costs out of control within 18 to 24 months.

Now think about what AI marketing tools actually accelerate. They accelerate the production of activation content: ad copy, email sequences, social posts, landing page variations. They're brilliant at the tactical 40%. They do almost nothing for the strategic 60%.

This is the mechanism behind why AI can make your marketing faster, cheaper, and identical to everyone else's. The tools amplify tactical work while the strategic work sits untouched.

The Costly Signal You're Accidentally Erasing

Rory Sutherland, Vice Chairman of Ogilvy UK, has spent decades arguing that "the meaning and significance attached to something is in direct proportion to the expense with which it is communicated."

This is costly signaling theory. A FedEx letter is assumed important because who would spend eight dollars to send something trivial? A wedding invitation on gilt-edged card stock signals different importance than an email, even with identical information. The effort IS the message.

Now consider what AI did to your marketing.

Before AI, producing a monthly newsletter, three blog posts, and a video took 20 hours of genuine effort. That effort was visible in the output. The quality was sometimes uneven but clearly human. Your audience could sense that a real person wrote this, thought about this, cared about this.

After AI, the same output takes three hours. You use the other 17 to produce five times more content. The quality is consistent, polished, and identical in tone to everything your audience encounters that day. Because everyone's AI draws from the same training data, optimises for the same engagement patterns, and produces the same kind of competent-but-generic output.

Sutherland's insight cuts deep here. When marketing looks cheap to produce, it signals that the business didn't invest much in the message. And if the business didn't invest in the message, the audience unconsciously infers the business might not invest in the service either. Your marketing used to be expensive, and that was the point.

The DoubleTree hotel chain gives every guest a warm chocolate chip cookie at check-in. Sutherland notes that 14 years later, he still remembers it, while he can't name a single distinguishing feature from 95% of the hotels he's stayed in since. One memorable gesture outweighed thousands of adequate ones.

Your marketing faces the same choice. Twenty forgettable AI-generated posts, or two that someone actually remembers?

The Real Bottleneck AI Skipped Over

Mark Ritson teaches a three-step framework: diagnosis, strategy, then tactics.

Diagnosis is research. Understanding your market, your customers, your competitive position. Strategy is choosing who to target, how to position, and what to prioritise. Tactics are the execution: the ads, the content, the emails.

AI supercharged this exact problem. It made tactics essentially free. An SME can now produce in an afternoon what used to take a week. But diagnosis and strategy? Those require thinking, customer conversations, market understanding, and judgment. AI can create marketing, but almost nobody can edit it because editing requires the strategic judgment that comes from doing the diagnosis work first.

The difference between an SME producing 30 pieces of AI content a month and one producing 8 pieces informed by genuine customer insight is not volume. It's effectiveness. Binet and Field's data across 996 campaigns makes this unambiguous: the strategic input determines the outcome far more than the tactical volume.

What Spotify Actually Got Right (That Most Marketers Missed)

Go back to the Spotify story. The headline was "engineers stopped writing code." But the real story was what they started doing instead.

They didn't use the freed-up time to write MORE code through AI. They shifted to architecture decisions, product strategy, and quality review. The nature of their work changed. They moved upstream.

Azeem Azhar, writing in Exponential View, captured the broader pattern: as AI scales, the human role shifts from production to specification and judgment. At a hundred tokens a day, AI is a toy. At a million, it's a workflow. At a hundred million, it's a workforce. But at every level, someone still has to decide what the workforce should do.

For SME marketing, this means the question has changed. It's no longer "how do I produce more marketing?" AI solved that. The question is "how do I decide what marketing is worth producing?"

That question requires diagnosis. Customer research. Competitive analysis. Understanding the specific situations that trigger someone to search for your service. Building a compound marketing system where each piece of work makes the next piece more effective, rather than adding more disconnected pieces to the pile.

What This Means for Your Business

If you adopted AI marketing tools in the last year and feel busier than ever, you're not doing it wrong. You're doing exactly what the research predicts. The tools expanded your capacity and you filled it.

Here's how to break the pattern.

The businesses that will win with AI marketing aren't the ones producing the most. They're the ones who used AI to buy back time, then spent that time on the things AI genuinely cannot do: understanding their customers, making strategic choices, and creating marketing that's worth remembering.

Everyone else is just running faster on the same treadmill.

Further Reading

• AI Doesn't Reduce Work, It Intensifies It - The Harvard Business Review study on workload expansion in AI-augmented workplaces
• The Long and the Short of It: Binet & Field's Framework Explained - Overview of the IPA effectiveness research on brand vs activation budgeting
• Magnitudes of Intelligence - Azeem Azhar on how each order of magnitude of AI usage changes the nature of the work
• Mark Ritson: Marketers Who Skip Brand Research Are Doomed to Fail - Why diagnosis must come before strategy
• Rory Sutherland on Costly Signaling and Trust - Why the effort behind your marketing is itself a signal of quality