AI Marketing Industry Trends

The Buyer Journey Has Changed. Your Funnel Hasn't.

February 10, 2026
9 min read
The Buyer Journey Has Changed. Your Funnel Hasn't.

29% of B2B buyers now start their research through LLMs more often than Google. Over half of consumers already use AI for product discovery. 94% of buying groups have already ranked their preferred vendors before a salesperson ever picks up the phone.

The buyer journey has fundamentally changed. Most marketing funnels haven’t noticed yet.

Tahnee Perry — founder of A25 and creator of the AI at Work newsletter — laid this out clearly at Section’s AI Marketing Strategy Summit. Her framework for the AI-native buyer journey maps almost perfectly onto the copilot-to-CMO gap we’ve been exploring in this series.

This is Part 2 of our trilogy. Part 1 established that copilots make creation free but leave orchestration to the marketer. This post asks the harder question: what does orchestration even mean when the buyer journey no longer follows a funnel?

The Funnel Is Dead. Long Live the Spaghetti Bowl.

Perry describes the modern buyer journey as “spaghetti on a wall.” She’s not being hyperbolic.

6sense’s 2025 B2B Buyer Experience Report surveyed 4,510 recent B2B buyers and found:

  • Average buying committees: 10.1 people, each researching independently
  • 16 research interactions per person with the winning vendor — over 160 across a full committee
  • 94% of buying groups had ranked their preferred vendors before first contact
  • They purchased from their preliminary favorite 77% of the time

BCG has proposed replacing the funnel entirely with what they call an “influence map” — a framework built around four simultaneous behaviors: streaming, scrolling, searching, and shopping. These don’t happen in sequence. They happen simultaneously, overlapping, at every stage.

Gartner visualizes the B2B buying journey as a “spaghetti bowl.” Buyers move forward, backward, go in circles, or abandon the journey altogether. Then they come back three weeks later through a different channel.

Here’s why this matters for marketing teams: scheduled campaigns assume linear movement. A drip sequence that sends Email 1, then Email 2, then Email 3 assumes the buyer is moving forward. But the buyer might have already asked ChatGPT about your product before Email 1 arrived, compared you to three competitors on Perplexity during Email 2, and then forgotten about you entirely by Email 3 — only to rediscover you when an AI agent recommends you in a different context two weeks later.

A copilot helps you draft each email. It doesn’t know the buyer is already at step 7 when you’re sending step 2.

Your New Homepage Is a ChatGPT Response

The shift in how buyers research is staggering in both speed and scale.

ChatGPT went from 300 million to 800 million weekly active users in ten months. Perplexity processes 780 million queries per month, growing roughly 20% month-over-month. Gartner predicted traditional search engine volume would drop 25% by 2026 due to AI chatbots.

Whether the specific number lands at 25% is debatable. The direction is not.

G2’s 2025 Buyer Behavior Report found that 8 in 10 B2B decision-makers say AI search has changed how they conduct research, with 29% starting via LLMs more often than Google. Two-thirds prefer engaging with vendor salespeople only in later stages — a significant increase year-over-year.

Yotpo’s research shows 51% of consumers already use AI tools for product discovery — and brands optimized for AI see conversion rates 9x higher than traditional traffic.

The implication: your most important “homepage” might be a ChatGPT response you’ve never seen and can’t control.

Microsoft published a comprehensive playbook in January 2026 distinguishing two new disciplines:

  • AEO (Answer Engine Optimization): Optimizing content for AI assistants and agents
  • GEO (Generative Engine Optimization): Optimizing for generative AI search environments

Their framing captures the shift perfectly: “SEO is how you won in search. AEO and GEO are how you win the recommendation.”

We wrote about this from the agent perspective in A2A Marketing: When Your Next Customer Is an AI Agent. The buyer side of that equation is now confirmed by the data. Your content needs to be simultaneously optimized for humans who read it and AI systems that cite it.

Personalization Isn’t Optional Anymore

Perry’s third theme — personalization at scale — is where the spaghetti bowl meets reality.

When every buyer takes a different path, generic campaigns can’t keep up. BCG’s research says every buyer needs their own “influence map.” 6sense says buying committees are 10+ people, each taking a different path. The logical conclusion: you can’t serve them all with one campaign, or even five.

DoorDash demonstrates what this looks like in practice. They built a GenAI-powered homepage that creates personalized carousels for each user — using LLMs and order history to balance familiarity, affordability, and novelty. Their Zesty app lets users search with prompts like “cozy pasta spots on the Lower East Side with outdoor seating under $100.” They’ve integrated with ChatGPT so users can get recipe suggestions and build shoppable grocery lists without leaving the conversation.

This isn’t experimental. It’s production.

Now apply this to marketing. A marketer with a copilot can personalize for 3–5 segments. Maybe 10 if they’re ambitious. An AI CMO personalizes for hundreds of micro-segments across channels simultaneously, adjusting in real-time as buyer signals change.

This is the Jevons Paradox playing out again: when the cost of personalization drops to near-zero, the rational response is to personalize everything. The question isn’t whether to personalize — it’s whether you have the infrastructure to do it autonomously.

Driver, Not Passenger

Perry’s most practical framework is her “Driver vs. Passenger” metaphor for how marketers should relate to AI.

A passenger uses ChatGPT to draft a blog post when asked. A driver directs an AI system that autonomously discovers what content to create, for which audience, across which channels, optimized for both human engagement and AI citation.

This maps directly to the copilot-to-CMO progression:

Passenger (Copilot)Driver (AI CMO)
“Draft me a LinkedIn post”System identifies what to post based on audience signals
”Write an email for this segment”System personalizes emails for hundreds of micro-segments
”Analyze this dashboard”System monitors all channels and adjusts in real-time
Reacts to your instructionsReacts to the market

Perry describes how marketing roles are already evolving:

  • Content creators become “editors in chief” — overseeing AI output, not producing from scratch
  • SEO specialists become “discovery strategists” — optimizing for Google AND ChatGPT AND Perplexity
  • Campaign managers become “orchestration architects” — directing AI systems, not operating tools

This is exactly the pattern we saw in Anthropic’s own marketing team. They saved 100+ hours a month with Claude on content creation. But every workflow they automated was creation. None was orchestration. The copilot made them faster at drafting. The orchestration — distributing, scheduling, monitoring, adjusting — still required humans.

The driver doesn’t type prompts. The driver sets strategy and lets the AI CMO execute.

New Metrics for a New Journey

If the journey has changed, so must the measurements.

Perry argues for moving beyond clicks and impressions to metrics that reflect how buyers actually behave in the AI-native era:

Answer Coverage — How often does your brand appear in AI-generated answers for your category?

AI Share of Voice — What percentage of AI responses cite you versus competitors? Industry analysis suggests brands at 25-30% AI SoV show market leadership; below 10% means you’re effectively invisible to AI-assisted buyers.

Citation Quality — When AI mentions you, is the information accurate and current?

Intent-to-Mention Ratio — For how many relevant queries does AI include your brand?

HubSpot has launched an AI Share of Voice tool. Gartner predicts 1 in 5 purchases will be completed by an AI agent in 2026. McKinsey projects agentic commerce could orchestrate $900 billion to $1 trillion in U.S. B2C retail revenue by 2030.

These aren’t vanity metrics. They’re leading indicators of whether your brand exists in the buyer’s AI-assisted research process.

And here’s the key: monitoring these metrics is inherently an autonomous task. You can’t manually query ChatGPT, Perplexity, and Claude every day to check how they describe your brand across hundreds of relevant prompts. This requires persistent, always-on intelligence — not a copilot you invoke when you remember.

The Trilogy Arc

Part 1 established the copilot ceiling: Claude’s marketing plugin is powerful, but it stops at “draft.” Anthropic’s own team proved this — they automated creation, not orchestration.

This post shows why that ceiling matters more than most teams realize. The buyer journey has fundamentally changed. The funnel is a spaghetti bowl. Buyers research through AI before visiting your website. Personalization at scale is table stakes. And the metrics that matter — AI Share of Voice, Answer Coverage — require continuous autonomous monitoring.

The copilot ceiling isn’t just a product limitation. It’s a strategic gap between how tools work and how buyers behave.

But there’s a deeper question lurking behind all of this. If over half of consumers now discover products through AI, and buyers rank vendors before any human contact, and AI agents may complete 1 in 5 purchases by year-end — does brand loyalty even exist anymore?

Scott Galloway — whose Section platform hosted Perry’s talk — has argued at Cannes Lions that “the era of brand is over.” We disagree. But the argument is worth taking seriously.

That’s Part 3.

Lane handles the new buyer journey autonomously — discovering channels, orchestrating campaigns, and monitoring your AI visibility 24/7. See how it works →

References

#Buyer Journey #AEO #AI CMO #Personalization #AI Share of Voice #LLM-SEO
Share this article
Twitter LinkedIn
Back to all articles