I’m going to say what every marketer with a data-science background is thinking but is too afraid to say out loud: the marketing funnel is a lie. A comfortable, profitable, century-old lie.
It’s a neat, linear diagram that gives CMOs a sense of order in the chaotic universe of consumer behaviour. Awareness, Interest, Desire, Action. It’s clean. It’s logical. It fits perfectly onto a PowerPoint slide. The only problem is, it’s a complete fiction. It represents a Platonic ideal of a customer journey that has never truly existed, and in the era of agentic artificial intelligence, clinging to it is a strategic death sentence.
For years, we’ve all paid lip service to the 'messy middle,' that inconvenient loop of exploration and evaluation that spoils the clean lines of our funnels. We’ve tried to patch it with multi-touch attribution models and complex user journey maps. But we were just redecorating a condemned building. Agentic AI is the demolition ball, and it’s already swinging. The linear path to purchase is not just messy anymore; it has been irrevocably shattered.
The Industrial-Era Logic of a Dying Model
The AIDA model, the intellectual grandparent of our modern marketing funnel, was conceived by St. Elmo Lewis in 1898. Let that sink in. Our core strategic framework for navigating the digital marketplace is based on a concept developed before the invention of the radio, let alone the internet. It was designed for a world of passive consumption, where brands broadcast messages and hoped for the best.
The logic was sound for its time. You create awareness (a newspaper ad), generate interest (a compelling product feature), foster desire (an aspirational image), and drive action (a visit to a physical store). It was a model built for an industrial economy of mass production and mass media. Its persistence is a testament to our industry’s institutional inertia, not its effectiveness.
The funnel’s very structure creates organisational dysfunction. It splits marketing departments into silos. You have the ‘top-of-funnel’ team, obsessed with reach and impressions, buying expensive billboards or programmatic ads. Then you have the ‘bottom-of-funnel’ team, relentlessly optimising for conversion with discount codes and retargeting pixels. The vast, complex space in between, where decisions are actually made, becomes a no-man's land of mutual blame.
This is not a theoretical critique. At my agency, Creative Marketing Group, we inherited this fractured thinking from new clients all the time. We'd see huge budgets poured into brand awareness campaigns with no measurable impact on sales, while the performance marketing team was squeezed to deliver impossible short-term results. The funnel forces a false dichotomy between ‘brand’ and ‘performance,’ when the reality is that they are two sides of the same coin.
The ASOS Case: How Agentic AI Kills the Funnel
Here’s my contrarian take: the beautifully optimised, data-driven funnels of sophisticated e-commerce players are now their greatest liability. Take a UK success story like ASOS. They mastered the classic funnel. They create awareness on TikTok, nurture interest with their app, and drive action with a slick checkout process.
Now, enter an AI agent. Not a simple chatbot, but a true, autonomous agent tasked by a user with a specific goal: 'Find me a sustainable, blue, size 10 dress for a summer wedding. My budget is £150, and I need it delivered by Friday.'
This agent doesn't scroll TikTok for inspiration. It doesn’t get 'influenced.' It has no 'brand affinity' for ASOS. It operates on pure, cold, hard data. It will likely execute a series of actions in milliseconds: query multiple retail APIs, scan product feeds for SKUs matching the user's constraints (sustainability a key attribute), compare shipping times, and analyse pricing data. The agent isn't 'aware' of ASOS; it just sees a dataset. It isn't 'interested'; it's matching parameters. It has no 'desire'; it is executing a command.