The AI era demands a new marketing funnel

Your classic marketing funnel is rapidly evolving. In its place is something smarter, faster and infinitely more personal: an AI-driven marketing loop. For marketing leaders, it’s time to stop thinking in linear stages and start designing a dynamic, intelligent journey. Here’s what’s changing – and how to get ahead.
Key takeaways
- Traditional marketing funnels no longer reflect how buyers behave.
- AI/LLMs turn static journeys into adaptive, ongoing loops.
- The “Messy Middle” of customer decision-making is now heavily influenced by what LLMs (large language models) choose to cite and recommend.
- Marketing leaders must rethink structure, team roles and data use to stay competitive.
- The future of marketing is intelligent, multimodal and never final.
From funnel to flywheel: A new focus for AI marketing
“The marketing funnel isn’t broken – it’s just no longer a funnel”
— Martin Harris
For decades, marketers built a strategy around the AIDA funnel. It worked well when communication was one-way, media was fixed and attention was a prize to be captured.
AI and large language models are not just optimising the traditional funnel – they’re transforming it. The linear funnel falls short of capturing today’s dynamic customer journey. A more fitting model is the flywheel or an interconnected loop, where sustained engagement and customer experience generate continuous momentum for growth.

Popularised by HubSpot, the flywheel approach places the customer at the center, with satisfaction fueling referrals, repeat sales, and growth. While not an entirely new concept, it aligns far more closely with how AI has reshaped marketing, shifting the focus from “push” tactics to “pull” strategies powered by social proof and AI-driven answers across every stage of the journey.
- Funnels end, flywheels compound. The funnel assumes a customer journey ends at purchase. The flywheel emphasises what happens after – retention, advocacy, referrals,which in practice often generate more growth than net-new leads.
- AI and LLMs extend the journey. People don’t move neatly through awareness–interest–decision. Instead, they ask questions, seek validation, and engage in loops of discovery. AI accelerates this non-linear process by giving answers at any stage.
- Customers drive momentum. In a funnel, the marketer pushes prospects down a path. In a flywheel, customers themselves create the force that drives growth through; reviews, testimonials, and community, making it more resilient and scalable.
Stage | Traditional Funnel (Tactics) | AI/LLM Funnel (Capabilities) | AI influence |
Awareness (Attention) | Mass advertising, broadcast TV, static SEO blogs, general social posts, cold email lists | AI-personalised content, multimodal discovery tools, real-time behavioural targeting | Humans ask AI about needs, reducing search. AI proactively educates. |
Engagement (Interest/Desire) | Email drip campaigns, generic whitepapers/eBooks, gated assets, fixed webinars, static landing pages | Real-time personalisation, dynamic journeys based on micro-signals, human-like chatbots, intent modelling from natural language inputs | Humans prompt AI for solution: AI proactively dialogues, analysing needs |
Consideration | Human compares products, reviews on sites/apps | (similar real-time personalisation/decision stages above) | Humans ask AI to compare products: AI agents shop & auto-compare for humans |
Decision (Action/Purchase) | Form fills, fixed-path conversions, call centre scripts, one-size CTAs | Adaptive chatbots, predictive content delivery, continuous engagement loops, remarketing triggered by emotional tone | AI refers directly to websites. (Near future) AI agents buy on human request. |
In 2023, Gartner forecast that by 2026, 80% of marketing content would be created with generative AI. Looking at today’s trends, that prediction seems hard to dispute.
Old funnel vs. new flywheel approach: What’s changed?
1. Awareness (attention): The rise of the “Messy Middle” and GEO
The traditional linear funnel assumes a customer moves neatly from awareness to decision, but as Google’s research on “The Messy Middle” confirms, modern buyers bounce between “exploration” and “evaluation” stages in a continuous loop before making a purchase.
This chaotic journey is where consumers are gathering information, comparing options and being influenced by a constant stream of new content. The way we search is changing fast, and platforms like ChatGPT, TikTok, and Google’s own AI Mode are replacing the traditional “ten blue links” with synthesised, conversational answers and videos.
“Funnels force customers to behave. AI listens instead.”
This is the essence of generative engine optimisation (GEO), as written about in this blog from my colleague Jake, Tank’s SEO lead. You’re no longer optimising for a search engine’s crawler; you’re optimising for generative models as well. The goal isn’t just visibility, it’s being useful, trustworthy and clear enough to be included in the answer itself.
Now, AI helps us to:
- Predict intent from behaviour patterns to reach the right person at the right time.
- Auto-generate content variants based on persona and format, so a single blog idea becomes a dozen personalised posts.
- Use LLMs to surface questions customers didn’t even know they had, turning a search query into an interactive learning experience.
AI also guides the discovery process with increasingly personalised content, essentially driving the consumer journey rather than the consumer driving it themselves.
AI mode by Google: YouTube video
“The future of marketing is about the future of experiences.”
— Nikos Acuna, Partner and AI Global Practice Lead at Davis and Partners Worldwide.
2. Consideration (interest & desire): Adaptive content and the LLM’s evaluation
The usual way of creating interest and desire was through a sequential journey. B2B Marketers would drip-feed emails over weeks, offering generic gated assets like whitepapers or eBooks. This content was static and the path was the same for every lead, regardless of their specific needs. This model fails to account for the dynamic, on-the-go nature of the modern buyer journey, a reality underscored by McKinsey & Company’s recent State of the Consumer trends report 2025.
The report’s findings highlight how consumer behaviour is no longer a predictable path but a continuous process of evaluation and re-evaluation, heavily influenced by real-time digital interactions and evolving preferences.
Now, generative engine optimisation extends beyond awareness. Content optimised for generative models – like comprehensive guides and detailed product comparisons – becomes the source material for the AI’s synthesised answers. As buyers loop through the “Messy Middle,” your content needs to be readily accessible and dynamic.
So, what can dynamic content look like in the consideration phase?
- Real-time personalisation is possible. AI-driven personalisation engines such as Optimizely can dynamically adjust a website’s content or an email’s subject line based on a user’s latest interaction, creating a 1:1 conversation instead of a generic email.
- Dynamic journeys based on micro-signals, for example the time a user spends on a specific product page, outperform old-school segmentation.
- LLMs read emotional signals and sentiment from surveys, reviews, and queries, allowing you to diagnose desire rather than just assume it.
- AI can also be used to create “smarter” content, such as dynamic video, that is optimised for audience behaviour on platforms like TikTok and YouTube, as highlighted by RESLV.
McKinsey found that top-performing brands in personalisation grew revenue 40% faster than competitors.
How LLMs navigate the Messy Middle: New behavioural biases
The consumer’s journey now starts with an AI-generated summary, which acts as a new kind of trigger. An LLM’s recommendation is the ultimate shortcut. When a customer asks an LLM for advice, they are outsourcing the exploration and evaluation phases to the AI. Your goal is to ensure your brand is one of the sources the LLM confidently selects and cites.
The six key biases identified by Google still exist, but they are now filtered through the LLM’s “brain”. To win the citation, your content must satisfy both the human and the AI:
- Social proof: LLMs are trained on vast datasets. A brand with a high volume of positive social proof across platforms is a more confident citation.
- Authority bias: An LLM is more likely to cite a source from a reputable, high-domain-authority website or a recognised expert.
- Category mental shortcuts: For an LLM it’s all about structured data. Using schema markup (like Product, Review, or FAQ schema) on your website is the ultimate shortcut for an LLM to understand and categorise your business accurately.
- Framing effect: The way your content is presented affects the LLM’s answer. Clear, concise headings, well-organised FAQ sections, and a logical structure make it easier for the LLM to extract information and frame its answer in a way that benefits your brand.
Beyond text: The rise of multimodal marketing
The next wave of AI systems (like Google Lens, ChatGPT with vision, and advanced virtual assistants) can process more than just text. They understand images, video and audio. This means your content strategy must evolve to become truly multimodal.
- Images: Are your product images tagged with descriptive metadata? Is your branding consistent in every visual asset?
- Video: Is your video content transcribed, subtitled and tagged with keywords? An LLM can “watch” a video and summarise it, but only if it’s optimised to be understood by AI.
- Audio: Is your podcast or audio content transcribed? This makes it citable for LLMs that process audio, allowing your brand to be featured in spoken-word AI summaries.
3. Decision (action): The funnel’s new “end” is a loop
The traditional funnel sees a conversion as a final step – the end of the marketing journey. This approach, which has been critiqued by sources like McKinsey as being a “relic of the past,” completely ignores the post-purchase experience.
For brands that will succeed, the sale isn’t the finish line; it’s the starting gun for a new loop of engagement. The shift from “push marketing” to “pull marketing” means the power now lies with the consumer, who drives brand growth through advocacy and word-of-mouth.
Now, conversion is the start of a new loop where AI provides multiple new opportunities to win customers and keep them:
- AI enables continuous post-purchase engagement, surfacing new upsell or cross-sell triggers based on usage or feedback. The flywheel model is built on this principle, where satisfied customers become advocates, driving new business through referrals and positive reviews.
- Adaptive chatbots guide prospects through the decision phase with personalised information, automating the last-mile of the customer journey.
- Remarketing is no longer just about retargeting a page visit; it’s about delivering content based on a customer’s emotional tone or recent feedback.
“In 2025, retention isn’t just a metric – it’s a pivotal strategy. As acquisition costs climb and AI levels the playing field, brands that prioritise loyalty and lifetime value will outpace those still chasing cold leads.”
— Martin Harris
What should marketing leaders do now?
Customer journey audit
A customer journey audit maps how prospects and customers interact with your brand across every stage, from awareness to post-purchase. The goal is to identify friction, optimise touchpoints, and ensure your content serves both humans and LLMs.
Key steps:
- Map the journey: Define the stages customers move through (awareness, engagement, purchase, retention, advocacy).
- Identify touchpoints: List every interaction – ads, website visits, emails, reviews, support, and social.
- Analyse performance: Use data (bounce rates, conversions, cart abandonment) to spot friction and drop-offs.
- Add qualitative insights: Validate data with surveys, interviews, or direct walkthroughs.
- Prioritise fixes: Identify quick wins vs. deeper structural issues, focusing on the steps that matter most for retention and conversion.
A clear journey audit ensures you’re aligning brand messaging and experiences with the real paths customers take through the AI-shaped funnel.
Conduct a GEO (Generative Engine Optimisation) audit:
Conduct a GEO journey audit across all LLM’s to see what questions your audience is asking. Then, ensure your content is not only a great answer for humans but is also a trustworthy and easily digestible source for LLMs.
For an LLM to cite your business, it needs to quickly and confidently evaluate your brand. Ensure your website has all the necessary signals in one place:
- High-quality, authoritative content
- Strong social proof (reviews, testimonials)
- Strong brand presence
- Structured data and schema markup so the LLM can easily understand your products and services
- A clean, logical site structure that allows the AI to navigate easily
Media mix modelling
Beyond journey and geo audits, media mix modeling (MMM) unlocks the “what actually drives behavior” insight.
MMM helps quantify the contribution of each channel (digital and offline) in driving customer action across geographies. When paired with geo-level journey insights, MMM empowers you to:
- Evaluate which channels are working hardest in specific regions
- Optimize marketing spend by reallocating budget from low-impact channels
- Align strategy with the actual decision-making paths customers take in reality
Reimagine your team’s role:
AI isn’t a replacement; it’s a co-pilot. Pair creative marketers with AI architects. The creative team focuses on brand storytelling and emotional resonance, while the architects design the prompts, manage the data inputs, and build the automated workflows that allow your creative content to adapt and scale intelligently. This requires a new level of cross-functional collaboration that breaks down traditional silos between marketing, content and sales.
Embrace the ethics of AI marketing:
The more powerful the tools, the greater the responsibility in using them. As you harness AI to shape the customer journey, it’s essential to establish clear ethical standards.
- Transparency: Be clear to ensure a customer knows they are interacting with AI.
- Data privacy: Prioritise the secure handling of first-party data and be explicit about how it’s being used to personalise experiences.
- Authenticity: Use AI to enhance and scale your brand’s authentic voice, not to generate generic, inauthentic content.
Data is the fuel, not the destination:
Clean first-party data is no longer just for segmentation; it’s the raw material that trains and powers your AI. To own your data, invest in a customer data platform (CDP) to unify data silos and create a single source of truth for your AI – a necessary step to track the multi-session, multi-platform journey. Focus on data governance and privacy from the start.
“Your AI doesn’t need to be creative. It needs to be curious – and trained on your customers.”
— Martin Harris
The traditional marketing funnel has changed
What’s taking its place is something far more powerful: an intelligent journey, designed by you and powered by AI. The question isn’t whether you’ll adopt AI, but whether you can afford not to. Start with a single stage, own your data, and begin building the marketing engine of the future, today.
Related Posts