In my last post, I argued that content strategy is the first thing Agent-Based Marketing (the new ABM) disrupts – and the first thing it rewrites. At Cannes Lions just over a month ago, that argument played out in virtually every conversation I had with customers. What struck me most, though, wasn’t the content question – there was broad agreement on that front. The harder question was this: if buyers are now conducting most, if not all, of their research through AI tools or interfaces, how do we as marketers know how to respond to our audience? We don’t know what they want, where they are in their decision-making, or when they’re ready to engage. This new intent data is richer than ever – but it’s occurring somewhere we cannot observe. Working out how to find it, interpret it, and act on it may well be the defining marketing challenge of this moment.

That is what this post explores. Part 1 considered where buyers arrive. Part 2 looked at what they find when they get there. This post focuses on what we cannot see – and what marketers can do about it.

The most important signal we’ve ever had is one we can’t see

For 20 years, the B2B world has tracked buyer intent across every form of web engagement: clicks, page views, form fills, searches, and time spent on page. Our entire measurement, lead, and journey management strategy has been built around scoring models, attribution frameworks, journey stages, and sales handoffs. This has evolved considerably – becoming quite sophisticated precisely because our audience was always interacting on surfaces we could observe.

Prompts change everything. At Forrester’s B2B Summit, researchers noted that the average prompt is 15 to 23 words long – considerably more than the 3 to 4 words typical of a traditional Google search. A buyer typing 'What’s the best marketing platform for an enterprise team in the B2B / tech industry that needs to optimise visibility and engagement with humans and AI agents?' is conveying far more intent in a single prompt than they would through a simple Google search for “web visibility product” or a website visit.

This is the paradox of the moment. The signal has become dramatically richer. Yet it remains, in effect, invisible.

The traditional customer journey has never been linear – we know this. But it was at least observable. We could track our audience moving from a Google search to a blog post, then to the product page and a demo. By following their online journey, we could gauge where they were in their buying process. We could enrich our signals with third-party intent data. All of that remains possible today – but the most influential moments are now taking place outside our trackable touchpoints, within conversations held with ChatGPT, Gemini, Claude, etc. As one might expect, these conversations generate rich, invaluable intent data – yet on platforms that do not necessarily share it back.

Forrester projects that organic search traffic will fall by 50 to 75 percent, yet conversions are expected to grow – visitors who do arrive on owned property are three times more likely to convert. Marketers now face a significant attribution and measurement challenge. The middle of the journey – where buyers are evaluating and vetting vendors – is becoming opaque. More than that, it is no longer the human doing the work; it’s a machine (I will return to this point later in this post).

From buying stages to buyer mindsets

This is where our industry needs to reframe its thinking. The shift is from attempting to determine buying stages to genuinely understanding the mindset of a buyer.

For decades, we have mapped buyers to stages: discover, explore, evaluate, buy, use, renew. Despite knowing that this journey is rarely linear, we have continued to score and qualify leads or accounts against this model. The entire purpose of developing a nurture programme or an engagement model is to move buyers from one stage to the next.

Yet when today’s buyers engage with LLMs, they are not progressing through defined stages. Rather, they are revealing their mindset – the specific problem they are trying to solve or the decision they are working towards. A stage tells you where a buyer sits on your map; a mindset tells you what they are genuinely trying to accomplish.

This is intent – though not in its traditional sense. It is conversational and expressed in natural language. Observed and analysed effectively, it can reveal not only buying intent or journey stage, but the buyer’s role, their business problem, their industry, and how deeply they understand the solution they are seeking. This is what rich intent looks like in the agentic era, and it can reveal the buyer’s mindset.

The distinction matters. The ‘validate my shortlist’ mindset and the ‘prove ROI’ mindset would both be placed in the ‘evaluate’ (mid/late funnel) stage. Yet in reality, each buyer needs something different – one needs a review, the other needs proof. Treating them in the same way means solving the wrong problem for at least one of them.

What this means for scoring, qualification, and handoffs

Scoring models need to adapt: The models we use to capture buyer journeys remain valid, and the buying groups we have built will not disappear. What we are currently missing, however, is that these models score only a fraction of the activity across the surfaces we own. Moreover, our buying group frameworks apply only to human participants – a fraction of what genuinely constitutes today’s buying group.

Intent data needs to expand: Our intent data has not lost its value. We should continue to rely on it as a basis for understanding top-of-funnel intent. That said, many of the providers we depend on are facing their own version of this disruption: their models rely on web traffic datasets that track less of the journey than before. We need to challenge ourselves – and the vendors we work with – to redefine ‘intent data’ in the agentic era.

The handoffs: In a B2B purchase cycle, buyers have typically completed 90% of their journey and decision-making before they ever speak to a sales representative. We are moving even closer to the zero-click purchasing model in B2B – closer to the consumer model than ever before. As the discovery and evaluation stages play out within AI tools, the handoff to sales looks fundamentally different. It is no longer a single trigger to pass leads to sales, but a gradual, orchestrated transition.

And finally, attribution: Avoid placing too much weight on channel attribution in its traditional sense. Every channel now plays a dual role: it is both an acquisition and engagement vehicle and a contributor to your brand visibility. If you optimise channels for a predefined journey or single outcomes, you risk optimising your organisation out of its own brand visibility. That’s a thread I’ll pick up further in Part 4.

When the buyer is a machine

In today’s buying groups, the first vendor interaction frequently occurs machine-to-machine. Agents are evaluating compliance requirements, benchmarking performance, prioritising technology by use case, and modelling cost against value. This is where we stand today.

Tomorrow – and indeed today – will grow even more complex. The machine will not simply prioritise and vet products; it will become the actual buyer, even in the B2B world.

We are not fully there yet – but we are not far from this reality. Teams that shift their thinking and begin to understand machine-to-machine signals now will gain a meaningful advantage over their competitors.

How we’re approaching this at Adobe

Our engagement model is evolving too. We are rethinking how we capture signal, score intent, and orchestrate the transition from marketing to sales – and two announcements we made at Cannes Lions are central to how we are doing it.

Adobe Brand Visibility, announced at Cannes, is a single solution for understanding and optimising your brand presence across both human and AI-driven experiences. It brings together Semrush’s data – including the world’s largest database of search and prompt behaviour – with execution capabilities, giving marketers clear visibility into how their brand appears across AI-driven and traditional surfaces. The signals that matter most aren’t the ones we’re buying. They’re the ones we’re earning.

Adobe CX Enterprise Coworker, which became generally available at Cannes, is our answer to the handover problem. You define the outcome. Coworker builds the workflow, pulls in the right data, activates the right specialised agents, and monitors performance against the goal in real time.

What comes next?

The new agentic signal is difficult to capture. The most impactful moments of the customer journey are invisible. Scoring models and qualification thresholds account for only a portion of what’s happening. The buying group now includes both humans and agents. For B2B marketers, visibility looks very different in today’s world – and that’s where your brand comes in.

In Part 4, I’ll argue that brand may be the most underrated asset in the agentic era.

This is Part 3 of a 5-part series on Agent-Based Marketing:

Part 1 – The New ABM: Why the most important acronym in B2B marketing just got rewritten

Part 2 – The New ABM: Why content strategy needs a ground-up rethink, not an SEO upgrade

Part 3 – The New ABM: Prompts, Not Clicks: Rethinking signal, intent, and the customer journey (you are here)

Part 4 – The Great Equaliser? Why brand matters more in an agentic world

Part 5 – Building the Agentic Marketing Org: Operating model, mindset, and the work ahead

https://business.adobe.com/fragments/resources/cards/thank-you-collections/brand-visibility