You Automated the Wrong Half of Marketing

Here's what most businesses do with AI marketing tools: they use a $200/month subscription to write social media captions that used to take 20 minutes.

That's like buying a Formula 1 car to drive to the shops.

The real capability isn't generating words faster. It's doing the strategic work that most SMEs never get around to doing at all. Market research. Competitor analysis. Customer language mining. Campaign planning. The stuff that takes 10, 20, 30 hours. The stuff that actually determines whether your marketing works or just looks busy.

The implication for your marketing isn't "AI can write your ads." You already knew that. The implication is: AI can do the strategic homework that most small businesses skip entirely. And that homework is where the real money is.

The Maths of When AI Actually Saves You Time

Wharton professor Ethan Mollick recently outlined what he calls the delegation equation for AI work. It comes down to three variables:

The trade-off is straightforward. You're deciding whether to do the work yourself or pay the overhead of directing and checking AI, potentially multiple times. The higher the probability of success, the fewer attempts you need. The longer the human baseline time, the more time you save per successful attempt.

Here's where it gets interesting. Look at where most businesses actually use AI for marketing versus where the equation says the biggest gains are:

A social post takes 20 minutes to write manually. Even if AI does it perfectly, you've saved maybe 15 minutes. Do that 50 times a month and you've saved about 12 hours. Not bad.

Competitive positioning research takes 15-20 hours. If AI delivers acceptable output on the first attempt, you've saved nearly 15 hours in a single session. And the output directly shapes every ad, every landing page, and every campaign you run for the next 12 months.

Why Strategy Keeps Beating Execution (And AI Hasn't Changed That)

If the maths is so clearly in favour of strategic AI use, why does almost every business default to execution?

Les Binet and Peter Field analysed nearly 1,000 advertising effectiveness case studies from the IPA Databank and reached a conclusion that remains one of the most robust findings in marketing science: the optimal split is roughly 60% brand-building strategy and 40% sales activation.

Campaigns near the 60/40 split outperformed on every meaningful metric: short-term sales, sustained market share, profit margins, and pricing power. Brands that pushed past 70% activation showed short-term gains but long-term decline. The effect is non-linear. Brand investment increases the efficiency of activation. Spend on activation in a high-brand-equity environment converts more efficiently than in a low-brand-equity environment.

Now look at how businesses allocate their AI effort. By some estimates, over 80% of AI marketing usage goes toward execution and activation: writing ads, scheduling posts, generating creative variants, optimising bids.

The Overconfidence Tax

There's a deeper reason businesses skip the research. Daniel Kahneman and Amos Tversky's decades of work on overconfidence bias found that professionals systematically overrate the quality of their own judgment, particularly in domains where they have experience.

You've been running your business for years. You know your customers. You know what they want. You know why they buy.

Except you probably don't. Not as well as you think.

Byron Sharp's research at the Ehrenberg-Bass Institute demonstrates why. If you ask most business owners to describe their ideal customer, they'll describe their best existing customer. But growth doesn't come from getting your best customers to buy more. It comes overwhelmingly from acquiring new and light buyers who don't look, think, or search like your loyal regulars. Sharp's data across 130+ brands in 13+ product categories is unambiguous: penetration drives growth, not loyalty.

Understanding those buyers, the ones you haven't met yet, requires research. Research that takes hours and hours, which is exactly why most SMEs never do it. They assume they already know the answer and skip straight to writing ads.

AI hasn't changed this behaviour. It's made it worse. When execution is fast and cheap, businesses produce even more marketing without ever questioning whether the strategy behind it is sound. We wrote about this in our piece on why AI can write your ads but can't tell you if they're any good.

Every ad campaign built on untested assumptions about your customer is paying the overconfidence tax. Every landing page designed around what you think matters, rather than what research shows matters, is paying it. Every month where your strategy stays the same because nobody's checked whether the market has shifted is paying it.

What Strategic AI Research Actually Looks Like

Rory Sutherland argues that most business problems are perception problems, not reality problems. His famous example: engineers spent £6 billion building new tracks to shorten the Eurostar journey by 40 minutes. Sutherland proposed installing Wi-Fi so passengers could work during the trip. The perception of wasted time was the problem, not the physical duration.

The marketing equivalent is everywhere. Businesses spend thousands optimising ad copy when the actual problem is that their landing page asks for too much information too early. Or their offer doesn't address the real objection. Or their competitors have framed the category in a way that makes them the default choice.

Finding these perception gaps used to take weeks. AI can do a version of it in hours. Here's what that looks like in practice:

What This Means for Your Business

The shift isn't complicated. It's a reordering of priorities.

Before you ask AI to write a single ad, use it to research:

• Who actually buys in your category (not who you think buys). What do they search? What language do they use? What are they worried about before they pick up the phone?
• What your competitors promise and where the gaps are. What are customers complaining about in reviews that nobody in your industry is addressing?
• Which buying triggers you're missing. Map every situation that causes someone to need your service. How many of those moments does your current marketing address?
• How your offer compares to alternatives. Not just on price. On perceived risk, perceived effort, and perceived confidence. Is your offer genuinely better, or just your ads?

Then let AI write the ads. It'll do a better job because it's working from real insight, not guesswork.

Mollick's equation predicts this approach will save you time. Binet and Field's data predicts it will produce better results. Sharp's research predicts it will unlock growth you're currently missing.

The businesses pulling ahead with AI marketing in 2026 aren't the ones generating the most content. They're the ones who used AI to understand their market first and then aimed their execution at the right problems.

You automated the easy half. The hard half is where the returns are.

Further Reading

• GDPval: Evaluating AI Model Performance on Real-World Tasks - OpenAI's benchmark of AI versus human expert performance across 44 industries
• Management as AI Superpower - Ethan Mollick on why delegation skills are the new competitive advantage in the agentic era
• AI Commoditises Marketing Execution and Elevates Judgment - MarTech on why strategic thinking is the new scarce skill
• The Long and the Short of It: Binet & Field's Framework Explained - The IPA effectiveness research behind the 60/40 rule
• How Byron Sharp's "How Brands Grow" Reshaped Market Research - Why understanding buyers matters more than persuading them