The 30-Minute Kitchen.

For about ten years I held a quiet, mildly embarrassing belief: I am a bad cook outside my own kitchen.

You probably know the version of this feeling. You rent a holiday home, you walk into an unfamiliar kitchen, and the entire instinct that normally carries you through a Thursday dinner just evaporates. The knives are dull. The stove has an agonising thermal lag. The prep space is cramped. The salt is hidden behind a cabinet door someone optimised for show, not for use. You spend the whole evening fighting the room instead of cooking in it, and you end up serving a meal that tastes vaguely like compromise.

I assumed this was a law of physics. Until one weekend, a few summers ago, I watched a friend disprove it inside two hours.

We were renting a slightly chaotic holiday house with twenty people in it. The kitchen was the standard rental kitchen — random pans, blunt knives, a stove that ran cold. I had already started low-key panicking about how dinner was going to work. My friend walked in, poured himself a glass of wine, and started cooking. Two hours later he served a four-course meal to twenty demanding adults. He never raised his voice, never rushed, and was finishing his second glass of wine when he plated dessert. The whole thing looked accidentally graceful.

When the kitchen was empty I asked him, in the genuine confusion of someone watching a magic trick: "How did you just do that in a kitchen that isn't yours?"

He smiled at me like I had asked a slightly silly question. He said: "I changed the kitchen. It took me less than thirty minutes."

I have thought about that sentence, in a slightly absurd way, every time I have walked into a CMO's office for the last three years.

This is where I get the most pushback in the room, and where I think the research is most useful. So let me put the numbers on the table first, then the framework.

The "we need a transformation" instinct fails empirically. McKinsey, Bain, and BCG converge on the same number: somewhere between 70% and 88% of enterprise digital transformations do not deliver against their stated objectives. The collective global cost of those failures sits at roughly $2.3 trillion a year against a projected $3.4 trillion in spend. Inside the failures, 70% trace back not to bad technology but to poor adoption — the new system arrives, 63% of employees abandon it within months, and the organisation quietly reverts to spreadsheets. Sixty-nine percent of workers describe their last enterprise change as a negative experience. That last number is the cultural debt that makes the next attempt even harder.

The agency-review version of the same instinct is roughly as expensive. A formal creative or media agency review costs the client-side brand an average of $408,500 in direct and indirect time, and the incumbent agency another $406,090 in defence. The total ecosystem cost of a single pitch is over $1.2 million and a full quarter of senior attention. Then the winning agency enters a six-month onboarding ramp — what the operational research literature calls the "Valley of Despair" — during which campaigns are paused, data pipelines are rebuilt, and client skepticism peaks at month three. Most of that performance dip wipes out the savings the pitch was supposed to deliver.

Meanwhile the actual MarTech utilization rate in enterprise marketing organisations, per Gartner's 2025 Marketing Technology Survey, sits at 49%. More than half of the features people already paid for are sitting dormant. Only 15% of organisations are classified as high-performing on their existing stack. And 50% of enterprises lack the basic data and infrastructure readiness to deploy the AI agents they have already bought. The board is being asked to fund a new kitchen while the existing one is operating at half its installed capacity.

So what are the three knives?

Data Practice. Stop trying to collect every byte of unnecessary behavioural data just because it is technically available. The competitive edge is no longer the size of your data lake. It is whether your clean, consented first-party data connects responsibly to your execution layers without a human in the middle. Most of the AI underperformance I see inside enterprise marketing teams traces back to this: the agent is being asked to make decisions on data it cannot actually see, because that data is stuck three platforms upstream behind a manual CSV export. Fix the plumbing before you buy another platform. The same Gartner survey shows that the "infrastructure deficit" — not the model quality — is what causes 45% of vendor-offered AI agents to underperform.

Creative Practice. Stop treating creative production as a manual assembly line of single-use assets, briefed one campaign at a time. Generative tooling, used inside a real brand-safety frame and not as a free-for-all, is now genuinely good enough to build deep, modular asset libraries that human creatives direct and curate rather than hand-build. Fifty-two percent of the AI agent deployments in marketing right now are aimed at exactly this — content and asset production — and they are the deployments most likely to pay back inside a single quarter, because they remove the bottleneck that throttles everything downstream. The agency review you were about to run, in many cases, is asking the wrong question. The issue is rarely that the ideas are stale. The issue is that the supply chain between the idea and the live ad unit is broken.

Media Practice. Stop pulling optimisation levers that the platforms now pull a thousand times a second whether you do or not. The job of a human media strategist in 2026 is to set the right objectives, define the guardrails inside which the auction is allowed to operate, and trust the system inside those guardrails. The teams I see winning are the ones who realised the algorithm is now the junior buyer; their senior people moved one layer up the stack, into strategy, audience hypothesis, and measurement design. The teams I see losing are the ones still convinced their best media planner manually beating Smart Bidding is a competitive advantage. It is, very consistently, not.

Those three changes — data plumbing, creative supply chain, media operating model — are the three knives. None of them require a new agency. None of them require a new platform. All of them can be started inside the existing budget, by the existing team, this quarter.

The research backs the operating choice. Organisations that pursue continuous, in-place tuning of their data, creative, and media practices achieve success rates roughly 5.3 times higher than those pursuing technology-first rebuilds. Same teams. Same budget envelope. Completely different return profile.