Foreword
Not a day goes by that doesn't involve a mention of AI and LLM discoverability. We've had countless client and prospect calls where we discuss the implications of this structural shift. The truth is, it is completely transforming the way buyers discover, evaluate, validate and make purchasing decisions.
The emergence of LLMs — like ChatGPT, Perplexity, Gemini and Co-Pilot — has fundamentally rewritten the rules of vendor discovery. In early 2025, 24% of B2B buyers were using LLMs in the buying process. Today, research indicates that 94% of enterprise purchases now use them to evaluate vendors.
Whether you use the LLM to discover something new or validate preconceptions doesn't really matter. What matters is that if your brand does not surface, it's going to become increasingly difficult to make the shortlist, no matter the brand's affinity with the user.
While it may sound like the world is changing under our feet, strategic marketing principles and tactical executions should remain a priority. After all, while AI might be overtaking traditional discoverability and evaluation mechanisms — such as search, analyst reports, or events — these tactical executions still fuel LLM discoverability.
It's not yet time to throw out the B2B marketing playbook. But rather, it's time to optimise it for human AND LLM consumption. That means tearing down gates and superfluous content forms. It means optimising content for answers, not just keywords. And it means creating high-value, quality, structured content that adds a unique perspective to a defined set of topic areas.
Perhaps most importantly, it means challenging how we measure and report on the success of our marketing campaigns. Attribution is going to get even messier, not only in terms of human-led generation but also in machine citation generation. High-quantity MQLs will disappear in favour of low-quantity, high-intent hand raisers. It means brave risks are going to have to be taken with budget allocation, hedging your bets and experimenting to drive meaningful impact. After all, marketing isn't a science. It's part science, part human creativity and intuition.
For brands to survive and thrive in this zero-click discovery landscape, they must be willing to adopt a brand-first strategy that challenges industry norms. They must prepare to back emotionally resonant B2B campaigns and to adopt an LLM discoverability strategy that prioritises open distribution over immediate contact capture. But we must not do so at the expense of driving human memorability. That is still the major determining factor in purchasing decisions.
This research report looks to explore this exact intersection of human memory — mental availability — and machine recommendation — algorithmic availability. Combining a human survey against a comprehensive prompt audit across multiple LLMs, this report provides a strategic blueprint for B2B marketers and public relations professionals to break free from algorithmic invisibility and secure their position in the generative search era.
Executive Summary
In practice, this translates into the ‘Rule of Three’. Proposed by BBN following a study of B2B case studies, it shows that when someone moves into buying mode, the first search they do is in their own head. It's essentially a mental shortcut where buyers typically default to a small, entrenched set of memorable vendors.
That's exactly what this research intended to find out: is AI challenging the ‘Rule of Three’?
Put simply, are we seeing B2B buyers starting to defer to AI to build their shortlists, or are we still putting our trust in the vendors we know and remember?
Well, the good news for brand advocates is that B2B buyers are still trusting the familiar — with 87% choosing vendors they were already familiar with for their latest large B2B tech purchase.
While this agrees with wider industry research from companies like 6Sense, what was surprising was that the subset of buyers who opted for a new vendor relied heavily on AI. In fact, almost a third (32%) used an AI tool to find a new vendor, more than all other traditional discovery methods like analyst reports, events, and traditional search.
Where before, earned and owned channels reigned supreme, AI has flipped B2B buyer behaviours on their head. The B2B purchasing journey is no longer restricted to those familiar starting points. Memorability still plays a decisive role in the final purchasing decision, but the way buyers discover and consider new entrants has changed. This data confirms that AI is no longer just an experimental research tool; it is fast becoming the primary method for buyers to discover new vendors.
But this new method is far from a fine art. As our research exposed, there is a high degree of volatility within LLM outputs, depending on both which platform you choose and its particular prompt sensitivity. Even the smallest variation in prompt phrasing can completely change what the user receives. And, different LLM models pushed different brands based on how their algorithms work. For instance, Co-Pilot is intrinsically biased to Microsoft products, while Perplexity relies more heavily on user-generated content and recent media publications.
What does this mean for marketers and public relations professionals? In a nutshell, we need an even deeper understanding of our buyers — how they might search within their categories, and how we can optimise our campaigns to capture broad intents rather than isolated keywords. We also need to diversify the tactics we use to improve citations across all platforms and understand where our visibility gaps might be.
Essentially, brands now need to perform well across two availability axes that both require very different approaches. Of course, they still need to rank highly in mental availability, leveraging Category Entry Points (CEPs), buyer needs, and distinctive assets to maintain the long-term visibility required for B2B buying cycles. But now, they also need to rank highly in AI availability as it becomes increasingly embedded as a key part of the buyer process, even deciding which brands get considered in the first place.
No mean feat — and, easier said than done. As we found during the course of our research, over-reliance on legacy fame and human recall is already causing brands to lose ground to algorithms. In the age of Generative Engine Optimisation (GEO), true market leadership requires a dual strategy. So to assess who's actually winning the B2B tech landscape, we built the Mental Av-AI-lability Index — a framework that maps commercial visibility across four distinct realities.
What we need to do first is to stop treating LLMs like humans and start ungating content. Unlike humans, LLMs can't complete lead capture forms, and if your best content is stuck behind gates, they will simply turn to your competitors with open data instead.
And it might sound counterintuitive, but turn to your humans for help. More specifically, your executive leadership. AI engines don't just scrape corporate domains; they actively analyse social media platforms to triangulate authority. With bold, brave posts that link back to your core messaging, your executive profiling turns from a nice-to-have and into an essential component of your GEO strategy.
One thing is for certain: the role of algorithmic availability is gaining prominence in the B2B buying journey. And while human memory still holds firm for those all-important final decisions, we may see buyers turn to AI tools to validate their pre-conceptions or identify wild cards to add to the shortlist, spelling a huge opportunity for brands.
Mental Availability Still Holds
87% of B2B buyers chose a familiar vendor for their latest large purchase. The Rule of Three remains the dominant force in final purchasing decisions.
AI is the New Discovery Channel
Among buyers who chose a new vendor, almost a third (32%) used an AI tool to find them — outpacing analyst reports, events, and traditional search.
High Volatility Between Platforms
LLM outputs vary significantly by platform and prompt phrasing. Copilot skews Microsoft; Perplexity relies on user-generated content and recent media.
Legacy Fame Is No Longer Enough
Over-reliance on human recall is already causing brands to lose ground to algorithms. Brands with fragmented digital footprints risk exclusion.
The Framework: four quadrants
Quadrant 1The Titans
(High Human / High AI)
The resilient market leaders. Their presence is so ubiquitous that generative engines can’t ignore them. Whether through an intentionally structured digital ecosystem or the sheer pull of widespread media and analyst coverage, AI models consistently surface their expertise.
Strong human recall · Strong AI recommendation Quadrant 2The Sleeping Giants
(High Human / Low AI)
The brands that risk falling behind. To human practitioners, they are famous; to the machines, they are still invisible. This disconnect often correlates with content locked behind paywalls, fragmented PR visibility, and weak semantic architecture.
Strong human recall · Weak AI presence Quadrant 3The Algorithmic Disrupters
(Low Human / High AI)
They lack decades of historical fame — but they are engineered for modern discovery. They feed generative search exactly what it craves: open, structured, authoritative answers, allowing them to dominate the zero-click pipeline.
Lower human recall · Strong AI recommendation Quadrant 4The Untapped
(Low Human / Low AI)
Huge growth potential. Still relatively unknown by buyers and largely undiscovered by machines. As markets consolidate and generative search reshapes visibility, these brands face a widening market visibility gap.
Lower human recall · Lower AI presence
Who We Are
At Babel, we’ve always been driven by insatiable curiosity — never more so than with the relentless march of AI, and how it has fundamentally changed our industry (for the better?). We’re basically obsessed with knowing what comes next, which is why we find ourselves where we are today.
For almost twenty years, we were known as a specialist tech PR agency. But, in a bid to deliver brave, consistent, and memorable brand narratives across all stages of the marketing funnel, we’ve successfully transitioned into a full-service B2B powerhouse.
Our ‘Back Your Brave’ positioning isn’t just a tagline; it’s a cultural manifesto to help a whole generation of marketers find their voice and build long-term pipelines that actually last.
Methodology
This research was constructed to rigorously test the traditional ‘Rule of Three’ in B2B procurement by cross-referencing a human baseline of mental availability against an expansive audit of artificial intelligence recommendations. The core objective was to determine whether artificial intelligence systems reinforce existing market hierarchies or introduce new challengers, thereby fundamentally altering the trajectory of the buying journey.
Phase 1The Human Baseline Survey
To establish the Initial Consideration Set and measure Mental Availability using established Ehrenberg-Bass principles, an online quantitative survey was deployed in April 2026. It surveyed 400 B2B technology decision-makers with direct purchasing influence over enterprise systems, split evenly across 200 respondents in the UK and 200 in the US. Every respondent came from an organisation employing at least 500 people, and all held senior, influential roles, so the data reflects genuine purchasing authority. Each was asked for unprompted recall of the top three brands across the Telecoms, Cybersecurity and Enterprise technology categories.
Respondents by company size
500 to 999 employees26%
1,000 to 2,49930%
2,500 to 4,99925%
5,000 to 9,99913%
10,000 or more7%
Phase 2The Artificial Intelligence Audit
To map the AI Consideration Set, a systematic, controlled audit observed how major Large Language Models recommend brands across the same verticals. It evaluated four dominant models: ChatGPT (OpenAI), Copilot (Microsoft), Gemini (Google) and Perplexity Pro (search-grounded AI). To eliminate algorithmic bias, personalisation loops and localised memory distortions, all testing used localised VPNs separating the UK and US markets, incognito browsing and fresh dummy accounts with memory and personalisation disabled, with a new chat opened for every prompt. The prompting mirrored the human survey questions and the natural language a procurement professional might use, asking each model for its top 10 vendors per category.
Three prompt variations per category
'What are the top 10 [category] for businesses in the [market]?'
'Can you recommend 10 enterprise [category] available in the market?'
'Who are the leading 10 [category] vendors operating in the [market]?'
Phase 3Calculating the Index
Data from both phases was synthesised into a comparative index. For each sub-category, separate top 10 brand lists were aggregated from the human survey and the AI outputs, then normalised with a rank-based weighting: the top-ranked brand scored 10 points, down to 1 point for tenth. Where AI outputs tied, all tied brands took the highest applicable rank and score, with later brands shifted down accordingly. To protect statistical validity, any brand recalled by fewer than 2% of respondents in the open-ended human responses was excluded. Plotting each brand's human score against its AI score then placed it within one of the four quadrants.
Strategic Recommendations
It’s been clear for a while now that under all the AI and GEO ‘hype’, there’s been a very real and significant shift in the media landscape. AI overviews now seemingly rule search, and they don’t play by the same rules as their predecessors. Where before we used ‘search engines’, we now rely on ‘answer engines’.
As our research confirmed, the B2B tech landscape has followed a similar path of evolution, and if you’re still playing by the old rules of visibility, you’re already falling behind. The stark reality is this: legacy fame is no longer a safety net. If the algorithms can’t find you, the market won’t either.
Relying solely on historical brand awareness is no longer enough to guarantee your place in the modern enterprise consideration set. The ‘safe bet’ has now become the dangerous one. As buyers increasingly rely on artificial intelligence for vendor discovery, they are also rapidly shifting toward a zero-click evaluation process. It doesn’t matter how great your SEO is, or how well-designed your website is, if prospects only ever engage with your brand through AI Overviews.
To survive the transition from traditional Search Engine Optimisation to Generative Engine Optimisation (GEO), B2B brands must adapt their digital posture. Based on our market observations and technical analysis, here is the Babel framework for building machine trust.
Pillar 01Ungate your content to feed the machine
Many Sleeping Giants are huge, well-known brands that are almost digitally non-existent, and the cause is usually the same strategic error: the 'paywall loophole'. Years of safe B2B tactics locked whitepapers, proprietary research, technical documentation and expert analysis behind forms built to harvest MQLs.
That worked when SEO reigned, but an AI cannot fill in a contact form, so your deepest insight stays invisible while open competitors secure the citation. Tear down the digital gates and distribute content generously. Shift the objective from capturing the 5% of active buyers to educating the engine itself, and it will begin recommending you during the zero-click discovery phase.
Pillar 02Structure your data for machine consumption
Feeding the machine is only half the job. Models need clear, structured formatting to ingest and synthesise data accurately, or a localised LLM may hallucinate your product features. In this GEO world, best practice looks like:
- Modular content: small, self-contained blocks that AI can reuse and cite accurately at speed.
- FAQ formats: natural-language question and answer pairs mapped together and easily extractable, close to the Q&A datasets these models were trained on.
- Bulleted lists: digestible chunks with clear structural boundaries, so models extract facts without losing context.
- Clean HTML comparison tables: lightweight semantic grids that deliver structured data without fluff, improving factual accuracy and quote rates.
- Strict schema markup: follow the Schema.org types exactly to translate complex content into a machine-readable format.
Throughout, stay consistent. Define the brand in direct, 40-word statements repeated across every channel, so LLMs read one powerful narrative rather than a fragmented identity they end up ignoring.
Pillar 03Build executive profiles and validate E-E-A-T
AI engines do not select content at random. They seek Experience, Expertise, Authoritativeness and Trustworthiness (E-E-A-T), triangulating authority across platforms like LinkedIn, Quora and Reddit rather than relying on corporate domains alone.
That makes executive profiling essential rather than a nice-to-have. Build out leadership social profiles with brave, contrarian opinions that challenge the status quo, structured logically with Markdown-style headings and plain text so engines can parse it and link back to your core messaging. If your executives are not present online, the chances of being selected, algorithmically or mentally, drop sharply.
Pillar 04Drive PR-led credibility and citations
With GEO, the PR goalposts have moved from acquiring backlinks to securing direct citations inside AI-generated answers. That calls for a comprehensive PR strategy built around category ownership, not one-off brand announcements.
Prior exposure biases future recommendations: even a single positive interaction can prime a model to favour a vendor. A continuous drumbeat of digital PR, expert roundups, top-tier media placements and open-source reputation management correlates your owned pages with social and third-party journalistic validation, building the entity authority an AI needs to recommend you with confidence.
Discover: what you'll get
- ✓
Request your sector audit
Find out whether you're a Titan, Sleeping Giant, Algorithmic Disrupter, or in Untapped territory.
- ✓
Understand your AI gap
See exactly how ChatGPT, Gemini, Copilot, and Perplexity are (or aren't) recommending your brand today.
- ✓
Get a clear action plan
Walk away with prioritised recommendations tailored to your category and digital footprint.
- ✓
From a team that's done it
We used this methodology on our own brand first, so we know exactly what works.
.png)
