The Best Ideas Don't Make Sense Yet

Key Takeaways

• The requirement that everything makes sense in advance fundamentally limits the size of the solution space — you only ever explore the part of the field that can already be explained, and that's the part everyone else can see too
• Creativity is search: the ability to expand the solution space and find answers that weren't visible inside the original framing of the problem
• The most underrated cultural problem in organisations isn't a shortage of ideas — it's a shortage of protection for ideas at the fragile stage
• Attribution is not causation: the last click takes the credit while the deeper work that created the demand gets filed under "brand"
• AI doesn't kill creativity; it makes average execution cheap and raises the price of unimaginative thinking
• The best creative work doesn't perform inside the existing market — it changes what the market wants

---

In a world optimised around what already exists, the rarest skill is becoming the willingness to follow an idea before it can explain itself.

There's a line I keep coming back to: the requirement that everything makes sense in advance fundamentally limits the size of the solution space.

It sounds abstract. It isn't. It explains why so much modern work is well-managed, data-informed, professionally aligned, and quietly lifeless.

We have built cultures that are very good at defending the known. The framework, the dashboard, the benchmark, the strategy deck, the attribution model. None of it is bad. But it sets a trap: when an idea has to justify itself before it has become anything, you only ever explore the part of the field that can already be explained. And the part that can already be explained is the part everyone else can see too.

Your competitors have the same data. The same category reports. The same case studies. The same AI tools. The same consensus about "what works." So when you insist that creative thinking move through a clean, sequential, rational process, you feel rigorous, but you may be walking the same road as everyone else, faster.

The danger isn't that process is wrong. It's that process can only take you to places it already knows how to reach.

Creativity is search

We talk about creativity in ways that make commercial people nervous. Inspiration. Genius. The muse. All useful, all slightly unhelpful. I've come to think of it in plainer terms: creativity is the ability to expand the solution space, to find answers that weren't visible inside the original framing of the problem.

A non-creative culture says, "Show me why this will work." A creative culture says, "Show me what this might reveal." The first demands certainty and protects the existing model. The second allows discovery and makes room for a better one.

That isn't a licence to abandon logic. Logic is brilliant at refining an idea: proving it, pricing it, packaging it, scaling it. It's just far less useful at the point of discovery, because real discovery usually requires movement before the map is complete.

The best workshop I ever sat in taught me one principle: explore the expanse of the thinking first, then refine. Most organisations do the exact opposite. An idea appears, barely alive, and before it has had any oxygen someone asks: Is this on brief? Can we prove it? Has anyone done this before? What would legal say? Sometimes those are the right questions. Asked too early, they don't sharpen the idea. They kill it.

That same workshop gave me a phrase I can't shake: nobody likes a poo in the pool. Every creative room has one, the person who arrives early with the objection, who mistakes negativity for intelligence, who says "just playing devil's advocate" and drains the energy from the water before anyone has had a chance to swim. They think they're being smart. But they're shitting in the pool.

The most underrated cultural problem in organisations isn't a shortage of ideas. It's a shortage of protection for ideas at the fragile stage. The first version of an idea is rarely the valuable one; the value tends to arrive on the third build, the strange collision, the moment someone says, "Wait, what if we took that seriously?" If the room rewards the fastest critic, you never get there.

"That doesn't make sense" usually means "that doesn't fit my model"

Every original idea begins outside the existing model. That's precisely why genuinely new ideas look wrong at the start. They don't have the language yet. They don't have the evidence yet. They don't fit the category yet. None of that makes them wrong. It often means they're early.

We like to pretend breakthroughs happen in straight lines: problem, hypothesis, process, result. Many don't. Darwin didn't deduce evolution from a strategy framework; he travelled, observed, collected, and only later did the pattern surface. The lesson isn't that accidents are a strategy. It's that breakthroughs demand the discipline to notice what the original plan wasn't looking for. Serendipity is not randomness. It's openness, rigour with wider peripheral vision.

We have built a mythology around measurement

Here's the uncomfortable part. Modern marketing has become brilliant at measuring what's easy to attribute and poor at understanding what actually created the demand. Performance marketing is treated as the rigorous end of the machine. To my mind, it's the art of taking credit for things whether or not it caused them.

A customer sees your brand for months. They hear about it from a friend, read an article, watch the founder speak, notice the product in culture. Trust builds slowly, invisibly, irrationally. Then they click a retargeting ad, and the dashboard says: we did that. Attribution is not causation. The last click takes the credit; the deeper work gets filed under "brand," which quietly becomes code for the things we know matter but struggle to prove — memory, distinctiveness, familiarity, desire, timing, belonging.

The data has been clear for over a decade. Analysing 996 campaigns across roughly 700 brands, Les Binet and Peter Field found that the most effective marketing splits about 60% to brand-building and 40% to short-term activation. Brand is what reaches the 95% of buyers who, at any given moment, aren't yet in the market to buy. And yet WARC's figures show the industry sprinting the other way: by 2024 nearly 69% of budgets were flowing to short-term performance, up from around 60% the year before. An entire discipline looked at the evidence and did the opposite, because the opposite produced a cleaner chart.

The cost of valuing only what can be cleanly measured. Means you slowly stop investing in the conditions that make people care in the first place. You harvest demand and call it growth. You optimise the tap and forget the reservoir.

AI raises the price of unimaginative thinking

I don't buy the idea that AI kills creativity. The opposite looks more likely to me. AI makes average execution cheap. It makes first drafts easy and "good enough" abundant, by early 2026, on Adobe's numbers, 87% of marketers had folded generative AI into a recurring workflow, up from around half two years earlier. But abundance is not originality. When everyone owns the same machines, the advantage shifts back to the quality of the human question. Not "can we generate more?" but "can we see differently?" AI does not remove the need for creativity. It raises the price of unimaginative thinking.

There's a second shift coming. AI won't just help us generate ideas; it will help us score them. Forecasting tools, synthetic audiences and prediction markets will increasingly sit around every creative decision. Some of that is genuinely useful. The risk is that "that doesn't make sense" simply becomes "the model gives this a low probability", the same old bias in a more objective-looking interface. Prediction markets are good at pricing consensus. Creativity is good at breaking it. A model can estimate demand. It cannot always see the work that creates desire, and the best creative work doesn't perform inside the existing market, it changes what the market wants.

Used badly, these tools narrow the solution space. Used well, they widen our peripheral vision: a strange phrase surfacing in niche communities, a behaviour spreading through group chats and comment threads, a market moving before the narrative catches up. The most valuable use of AI may not be prediction at all, but attention, not the judge of whether an idea deserves to exist, but the thing that says, "Something strange is happening over here." Spotting the outlier is the easy part. Knowing whether it's noise, novelty or the beginning of a movement is taste, and taste is still ours.

Search more of the map

In optimisation there's a name for the trap: the local optimum, the best answer reachable from where you already stand. It looks sensible, defensible, like progress. It can still be miles from the better answer, the one that needs a jump rather than a step. Any system that only refines gets stuck there. So do most companies.

This is why play matters, not because it's fluffy, but because it changes the search pattern. It lets you try things before they're justified, combine ideas that don't obviously belong together, and let the work reveal something the plan couldn't predict.

The future rarely arrives as a finished argument. It shows up as a hunch, a mistake, a tension, a half-formed thought that shouldn't work but somehow does. Someone still has to sharpen it, build it, test it, sell it, and take responsibility for it. But before any of that, someone has to be willing to follow it before it makes complete sense.

That is what creativity is. Not decoration. Not content. Not a workshop with sticky notes. It is the expansion of the possible, and in a world increasingly optimised around what already exists, it may be the most valuable skill we have.

AI will make the obvious cheaper. Our advantage will be to know what to do with the strange.

---

Frequently Asked Questions

Why do the best ideas not make sense at first?

Every original idea begins outside the existing model — that is precisely why genuinely new ideas look wrong at the start. They don't have the language yet, the evidence yet, or the category yet. None of that makes them wrong; it often means they're early. The requirement that everything makes sense in advance fundamentally limits the size of the solution space.

What is the 60/40 rule in marketing?

Analysing 996 campaigns across roughly 700 brands, Les Binet and Peter Field found that the most effective marketing splits about 60% to brand-building and 40% to short-term activation. Brand is what reaches the 95% of buyers who, at any given moment, aren't yet in the market to buy. Yet WARC's figures show the industry sprinting the other way: by 2024 nearly 69% of budgets were flowing to short-term performance, up from around 60% the year before.

Does AI kill creativity?

The opposite looks more likely. AI makes average execution cheap — it makes first drafts easy and "good enough" abundant, with 87% of marketers folding generative AI into a recurring workflow by early 2026, on Adobe's numbers. But abundance is not originality. When everyone owns the same machines, the advantage shifts back to the quality of the human question: AI raises the price of unimaginative thinking.

What is a local optimum and why does it trap companies?

In optimisation, the local optimum is the best answer reachable from where you already stand. It looks sensible, defensible, like progress — yet it can still be miles from the better answer, the one that needs a jump rather than a step. Any system that only refines gets stuck there, and so do most companies. Play matters because it changes the search pattern, letting you try things before they're justified.