Hello B2A!
The obvious objection, and we should deal with it before going further: modern models are multimodal. They parse images. Feed one a screenshot of your site and it will describe your layout, read your headline off the pixels, and tell you the logo is in the top left.
All true. All beside the point.
An agent can process your hero image. It cannot be moved by one. There is no taste to flatter, no mood to establish, no half-second of aesthetic pleasure that tilts a judgement. It has vision without appetite.
Atmosphere, sequencing, emotional resonance, the slow accumulation of trust across touchpoints. Every one of those instruments assumes a mind on the other end that can be nudged. Take away the nudge and most of the toolkit goes quiet.
What is left is unglamorous and unforgiving. Is the claim clear? Is the terminology right? Does anything else on the open web corroborate it?
Persuasion forgives vagueness. Retrieval does not. A page a human would charitably skim past as “a bit fluffy” is, to a machine, a page with nothing in it.
The assumption that breaks everything
In a note published on 16 June 2026, Forrester principal analyst Chuck Gahun set out what B2A actually requires. Agents do not browse your site or respond to marketing narratives, he writes. They retrieve, validate, and surface structured information to answer engines and consumer agents.
Then he drops the number that deserves far more attention than it is getting.
Fully 69% of AI decision-makers believe generative AI tools will always produce the same outputs, per Forrester’s State of AI Survey, 2025. They do not. Answer engines generate varied responses to each prompt. Forrester calls these non-deterministic experiences.
Sit with the size of that. Roughly seven in ten of the people making AI decisions are working from a mental model in which the machine is a lookup table, a thing you check once and consider handled. It is not. It is a system that reassembles your company out of scattered evidence, differently, every single time somebody asks.
You cannot QA that with a screenshot. You cannot sign it off in a launch meeting. And you certainly cannot fix it by publishing one more article about your industry.
If you do not shape the context, Gahun writes, agents will write a different brand narrative than you might have intended. They are not waiting for your approval.
The least glamorous phrase
Information architecture used to be a phase of website redesign projects. It is becoming the foundation for how you do business with machines.
That is a reclassification, and it costs money. Information architecture has always lived where budgets are thinnest: a few weeks of site mapping wedged between discovery and design, first thing compressed when the timeline slips, rarely revisited once the site is live. Nobody has ever been promoted for a good taxonomy.
Under B2A it becomes load-bearing. Gahun’s argument is that agents need straightforward information architecture in order to extract content and hold a conversation with a human buyer, and that your site should be fluent in your sector’s terminology precisely to minimise the need for agents to infer. Inference is where hallucination lives. Every gap you leave is a sentence something else writes for you.
Adobe’s numbers point the same way from a different angle. In its Q3 AI Traffic Trends Report (June 2026), Adobe Digital Insights found companies with the highest share of AI-referred traffic beat the lowest on homepage citation readability by 52%, with content and buying-guide pages 23% to 30% ahead.
That study covers US retail, travel and financial services rather than B2B, so take the direction rather than the decimal. But the direction is consistent. The winners are not the loudest sites.
Agentic Loyalty?
Gahun’s last idea has the longest tail. He calls it agentic loyalty.
Agents favour sources that prove reliable and consistent across multiple interactions and multiple sources. Not once but repeatedly. Your content competes against the constantly evolving training set of every rival you have, which means product data has to stay current, inaccurate claims about you scattered across the web have to be found and corrected, and your facts need to be easy to verify.
Content governance has stopped being hygiene. It is now a technical requirement for staying in the consideration set at all. For most firms that is not new tooling. It is new work.
If you are not an enterprise
Forrester’s respondents are enterprise decision-makers, and only 24% of the B2B ones say they plan to ensure their content is visible and authoritative in AI-powered search, per Forrester’s Marketing Survey, 2026, as reported by Forrester.
The translation down to a smaller firm is not the one people expect.
A large organisation carries scale as armour. It is covered in trade press, listed in directories, reviewed on comparison sites, and argued about in forums. When an agent assembles its story, there is plenty of corroborating material to work with, most of which the company never wrote.
A mid-size manufacturer, an interiors practice, a specialist legal or accounting firm has none of that cushion. The site is close to the entire evidence base.
It means the gaps do more damage. It also means the work is far more tractable. No eighteen-month platform migration stands between a focused firm and a clear, well-structured, terminologically honest presence. There is a decision, and a few weeks of unglamorous discipline.
Digital Maturity was never really about the tools a business owns. It is about whether the business can describe itself accurately. That was true when the audience was human.
It is merely less forgiving now that the audience has no eyes to charm.
Essentiate brings the discipline that makes a business legible, to buyers and to the machines increasingly standing between you and them. Our PIE Framework starts where this problem actually starts, with honest diagnosis rather than another content sprint, then builds and sustains Digital Engagement that holds up.
