In this episode of the Back to Brave podcast, Babel's Head of B2B Marketing, Ash Lockett is joined by Andrew White, CEO and co-founder of Sapio Research, to unpack Babel's newly released report, The Mental Av-AI-lability Index. Together, they explore the critical intersection between human brand recall and algorithmic recommendations within the B2B tech sector. The episode dives into the methodology behind the research, revealing how AI is rapidly shifting the B2B buying journey (notably that 32% of buyers who select a net-new vendor now use AI tools to find them - officially outpacing traditional search engines and analyst reports). Ash and Andrew also break down the four new commercial realities, identified in the report, into quadrants (Titans, Algorithmic Disruptors, Sleeping Giants, and The Untapped) and share actionable best practices for conducting impactful, unique brand and thought-leadership research.
Timestamps
0:00 - 1:21: Introduction to the podcast and guest Andrew White, detailing Sapio Research's focus on B2B and tech audience and brand research.
1:22 - 6:57: Ash introduces the Mental Availability Index and discusses the growing reliance on AI in B2B purchasing, highlighting the debate on whether brands should optimize for human memory or machine recommendation.
6:58 - 10:56: A breakdown of the research methodology, which included surveying 400 UK and US B2B tech decision-makers and auditing various LLMs to rank top brands.
11:08 - 15:05: Key findings on buyer habits: 87% of buyers still choose familiar vendors, but a staggering 32% of those choosing net-new vendors use AI tools for discovery.
15:06 - 22:33: Exploring the four brand quadrants: Titans, Algorithmic Disruptors, Sleeping Giants, and The Untapped. Ash and Andrew discuss how broad product portfolios from larger brands can cause "algorithmic dilution".
22:34 - 27:53: Best practices for conducting brand research, including the importance of using problem-specific questions over general ones, and the ideal frequency for tracking brand changes.
27:54 - 32:14: Tips for creating unique thought-leadership research by finding the intersection between audience interest and brand attributes. Andrew explains why specific AI centric use-cases perform better than generic AI surveys.
32:15 - 39:58: Advice on securing robust sample sizes (recommending a minimum of 400 for B2B) and ensuring audience representation. Andrew highlights common pitfalls to avoid, such as self-serving or rushed questionnaires, while Ash advocates for including "joker questions" to uncover unknown data points.
39:59 - 44:11: Final actionable tips on being bold with research methodologies, such as using monetary data extrapolations and maturity models to properly educate audiences.
Read the full Mental Av-AI-lability Index HERE.
www.babelpr.com
www.sapioresearch.com
Host: Ash Lockett
Guest: Andrew White, CEO and Co-Founder of Sapio Research
Recording Date: June 4, 2026
Ash Lockett:
Welcome to Babel's Back to Brave Podcast—the podcast for B2B marketing and communications professionals looking to make brave moves. I'm your host, Ash Lockett, and I'm joined today by Andrew White, CEO and co-founder of Sapio Research. Sapio is a full-service market research agency helping B2B and technology brands turn evidence into better decisions, more effective marketing, and strong market positioning.
Thanks for joining me today, Andrew. It's good to see you again. It seems like I've been seeing a lot of you recently, which is great. Before we dive in, just so our audience knows who you are and what you do, do you mind giving us a quick intro to yourself and Sapio Research?
Andrew White:
Thanks very much for having me, Ash. Absolutely. Sapio Research has been around for almost ten years now; it's our tenth anniversary this year. We're an agency that specialises in B2B and tech, working heavily in the cybersecurity, fintech, HR tech, and enterprise technology spaces. We tend to deliver work across three core segments: brand research, audience understanding, and content/thought-leadership research, utilising a variety of different methodologies to achieve that.
Ash Lockett:
Awesome. I think brand research, particularly within the B2B tech sector, is something that's probably not done enough, so it will be good to get your opinions on that a bit later in the episode.
Really, why you've joined me today is to discuss Babel's latest research. We are big advocates for conducting unique research in the market, and help a lot of our clients do just that, but we also practise what we preach. We've recently launched the Mental Av-AI-lability Index, which digs into the intersection of human memory and algorithmic recommendation.
Today, I want to look at the research results with you, but also look at research best practice, as it's becoming an increasingly important part of AI search strategies. We know Large Language Models (LLMs) in particular love information gain, and unique research is a core component for adding value and context to those digital conversations.
Ash Lockett:
Let's kick off and dive into our research. It would be remiss of me not to set the scene regarding why we conducted this and what we were looking at. We hosted an event the other week to talk through some of the key findings with a few members of our Brave Collective community. We found out that AI is now a deeply accepted part of the B2B buying process. When we asked the room how many people thought it was critical to B2B purchases, pretty much everyone raised their hand.
Analyst reports and industry data thoroughly agree. Forrester, for instance, states that 94% of B2B buyers rely on AI tools to validate decisions, and there is plenty of similar data out there. What this is leading to is every brand scrambling to "do AI." Not a week goes by where I haven't been asked by a client about how to get cited on Reddit or Wikipedia, Gartner is predicting explosive growth in AI-targeted PR, and everyone is talking about how to secure citations from LLMs.
At Babel, we wanted to see whether brands should actually be panicking; whether this is something we really need to start aligning all our content and marketing strategies around, or if it is mostly hype. The core component of this research was to find out whether it is more important to be humanly remembered or to be recommended by the AI tools that B2B buyers are now using on a near-daily basis.
The reality is that this question completely split the room. It shows we are at a tipping point where B2B marketing and comms professionals don't know whether they should keep optimising campaigns for humans (trusting that good marketing strategy will still ring true), or if we should pivot to optimising entirely for the machines to ensure we get recommended.
Before we dive into how we actually conducted the research, I wanted to ask, what were your initial thoughts on this research concept? Is it something you've seen done before, and what did you find interesting?
Andrew White:
The topic of AI search in terms of B2B buyers finding vendors is growing rapidly. We're seeing it ourselves; prospects are actually coming to Sapio via AI recommendations now, which is fantastic. Because of this shift, optimisation is becoming an incredibly important pillar of strategy for many brands.
While there have been a few baseline studies discovering this space conceptually, what has been missing is a look at the granular detail regarding what's happening to specific B2B brands. What's really fascinating about this piece of research is that it combines a traditional, vital marketing KPI—mental availability, looking at which brands first come to a decision-maker's mind in a specific sector, and pairs it directly with the algorithms. We examined how a range of different LLMs rank those exact same brands, charting the intersection between the two. Mapping out the brands that have massive success in both areas, versus those that are strong in one but weak in the other, is where this data really comes to life.
Ash Lockett:
For me, a lot of the research currently out there just asks general questions like, "Is AI used today?" which is becoming a bit of a 'no-brainer' question. That's what really excited us about this project — examining the real tension between brand fame and algorithmic availability. We honestly didn't know if it would produce a predictable result or if it would give us a definitive blueprint for how to move forward. There's always a bit of anxiety when you're doing something completely new. You find yourself thinking, "Please let the data be interesting!" That's how we knew we were being brave, because there was that distinct bit of trepidation going into it.
Let's walk through how we actually conducted the research. Could you provide an overview of how we went about it, who we interviewed, and how we handled the AI component? Because speaking to machines is obviously a bit different than speaking to humans.
Andrew White:
Absolutely. First, there was the primary survey portion of the research. We went out to 400 B2B technology decision-makers: 200 in the UK and 200 in the US. These individuals were vetted decision-makers operating across three core sectors: telecom, cybersecurity, and enterprise technology.
We asked these buyers an unprompted question to list the top brands that came to mind for specific service segments; for example, CRM providers in enterprise tech, or endpoint security vendors in cyber. We collated that data, identified the percentage of top-of-mind recall for each company, and ranked them 1 to 10 to establish our human 'Mental Availability' score.
For the LLM component, we tested a host of different leading AI platforms. We fed them a prompt mirroring the human questionnaire, asking them which brands they would recommend for those exact same service segments. Depending on the frequency and positioning of the mentions across our prompt testing, we ranked the AI responses 1 to 10 as well.
Finally, we took those two distinct top 10 rankings (the human recall and the AI recommendations) and plotted them across a two-by-two matrix to see exactly how they correlated.
Ash Lockett:
It's an important caveat for our listeners to note that this index specifically looked at the top ten market leader brands in each segment. While we've plotted their placements, there are hundreds of smaller, or niche companies underneath them that would populate a wider graph. But from a foundational standpoint, it gives us an incredibly clear directive view.
One of the most interesting nuances I found when running the prompts through the LLMs was how hyper-sensitive the machines are to phrasing. Just altering one word in the prompt, shifting from "who are the leading brands" to "who are the top brands" or "who would you recommend", produced completely different results. It was wild.
We also saw distinct biases across the platforms. Microsoft's Copilot, for instance, was deeply biased towards recommending Microsoft's own tools wherever applicable. But beyond the machine data, we also asked the human buyers about how these AI tools are actually impacting their real-world purchases. Let's dig into those stats, Andrew, because they really contextualise the index.
Andrew White:
When we looked at the human data surrounding their last significant B2B tech purchase, specifically looking at high-intent deals valued at £50,000 or more, we asked how important prior familiarity with the vendor was to their final decision. A massive 87% stated that they ultimately purchased from a vendor they were already familiar with before the process started.
However, we then isolated the remaining segment of the sample who had purchased from a brand they weren't previously familiar with, and asked how they discovered them. 32% stated they discovered the vendor using an AI tool. That was higher than any other channel, even beating out traditional search engines. It clearly demonstrates how rapidly AI search has rocketed up the ranks as a discovery tool for net-new technology providers.
Ash Lockett:
That is an insane statistic. Let's double-click on that for a second. On one hand, 87% of B2B tech buyers chose a brand they were already familiar with. That tells me that despite all the AI hype, brand fame, trust, and reputation still absolutely dominate the final decision-making process. Marketing and comms professionals shouldn't abandon traditional brand-building; the hype train hasn't completely left the station just yet.
But where it gets incredibly interesting is that out of the buyers who open-sourced their search to find a new vendor, 32% used AI. Extrapolated across the whole sample, that means roughly 1 in 20 major B2B purchasing decisions are now being driven directly by an AI recommendation. Considering generative AI tools were essentially in their infancy a few years ago, that is an astronomical leap.
It completely shook up the traditional discovery stack, beating out organic search engines (26%), analyst reports (19%), word of mouth (13%), and industry events at just 8%. Think about how much budget B2B brands pour into massive event footprints, yet AI is already proving more influential for net-new vendor discovery. It's a space we have to watch closely, even while remembering that brand familiarity still wins 87% of the time.
Ash Lockett:
So, we know humans trust familiarity, and algorithms are introducing wildcards. The Index allowed us to map these forces together across four distinct quadrants. Andrew, can you talk us through those four segments and how the brands grouped together?
Andrew White:
The matrix breaks down into four quadrants based on high or low human recall versus high or low AI recommendation:
Ash Lockett:
The concept of the Algorithmic Disruptor is fascinating because these brands are essentially acting as the procurement wildcard. If an executive runs a prompt and an LLM surfaces an innovative specialist they've never heard of, they're increasingly throwing them into the RFP process, just to do due diligence and test the market.
On the flip side, the Sleeping Giants are suffering from what I call "algorithmic dilution." Because these legacy companies have incredibly broad product portfolios, spanning dozens of sectors - they lack the hyper-focused, structured data nodes that LLMs love. An LLM wants to crawl the web and find absolute, niche specialisation for a specific problem. Humans use mental shortcuts to recall the biggest names they know. Whereas, LLMs do deep, structural crawls to find the most exact context.
Ash Lockett:
This brings us to how we fix this. To ensure you are discoverable by both humans and machines, you need to hone your messaging, positioning, and data structure. This is where rigorous research comes in.
Let's look at this from two perspectives: brand tracking research and thought-leadership research. Starting with brand tracking, if a marketer wants to conduct research to support their positioning, what are the best practices they should implement tomorrow to understand where they truly sit?
Andrew White:
There are two major areas I would recommend here. First, ensure your research questions aren't too general. If you're a cybersecurity vendor, you shouldn't just ask, "What cybersecurity brands are you aware of?" It's too broad and invites consumer bias. Instead, you need to test for true category entry points by anchoring questions to specific business problems. For example, ask: "If your company suffered a major data breach tomorrow, which specialist consultancies would you bring in to manage the incident response?"
Second, you must balance unprompted and prompted awareness. Unprompted awareness (like we used in this index), where respondents have to type the names from memory, reveals true top-of-mind mental availability. Prompted awareness, where you provide a checkbox list, helps you understand your wider market perception and what your brand is actually known for once recognised.
Ash Lockett:
I love that point. It perfectly aligns with the LinkedIn B2B Institute's work around Category Entry Points. Figure out the exact corporate trigger situations you want your brand to own, and build your tracking metrics around those scenarios rather than generic industry terms.
How often should B2B brands be running this type of brand tracking? In consumer marketing, brands track this monthly, but what's the sweet spot for B2B tech?
Andrew White:
For B2B tech, monthly tracking is generally unnecessary and economically inefficient. B2B perception shifts take much longer to manifest in the data. We typically recommend checking-in annually. An annual cadence gives your marketing, PR, and share-of-voice campaigns enough time to actually move the needle, providing a clean benchmark of what is working year-on-year.
Ash Lockett:
Let's pivot to the thought-leadership side of research - conducting studies designed to secure tier-one media coverage, build authority, and feed the LLMs with what they call "information gain". How do brands find that magical white space to create data that is genuinely unique?
Andrew White:
You have to find the exact intersection of three elements: what is top-of-mind for your audience right now, what hasn't already been done to death by competitors, and how it aligns with your brand's unique positioning.
To find the white space, you need to talk directly to your customers and prospects about their immediate, undocumented operational challenges. Combine that with rigorous desk research to see what data already exists. You're looking for the questions that industry professionals are actively asking, but simply don't have the hard data to answer yet.
Crucially, it has to fit your brand authority. If you're a fast-fashion retail brand, you shouldn't suddenly publish a dense report on corporate environmental sustainability, because it fundamentally jars with your business model. Find the sweet spot where the data solves an audience problem while organically driving your brand narrative.
Ash Lockett:
Because Sapio handles so much enterprise technology research, what are the most over-researched, fatigued topics that marketers should probably avoid right now?
Andrew White:
Unsurprisingly, it's general AI. But let me clarify, that doesn't mean you shouldn't touch AI as a topic. What's completely exhausted are generic, high-level surveys asking things like, "Is your business planning to adopt AI this year?" That adds absolutely zero value to the industry conversation.
The white space lies in hyper-specific operational impacts. For example, looking into how Agentic AI is transforming workflows within specific departments, or calculating the exact financial implications of deployment. Move away from generic sentiment and drill into specific, unmapped use cases.
Ash Lockett:
When executing these studies, a major logistical question is always sample size. To get a genuinely robust view that tier-one media and industry analysts will take seriously, what numbers should B2B marketers target?
Andrew White:
For B2B brand perception studies, we recommend a minimum of 400 respondents per market. You have to remember that only a fraction of your total respondents will be aware of certain challenger brands. If your total sample is too small, your data on specific brand attributes will degrade and lose statistical robustness.
For media-facing thought leadership, the minimum B2B sample size hovers around the same mark, though consumer studies require much higher numbers, like 2,000 nationally representative respondents.
However, the sheer size of the sample is actually less important than its representative balance. If you sample 2,000 people, but 80% of them are men aged over 45, you don't have a snapshot of the market, you have a snapshot of a single demographic. Your sample must accurately mirror the actual demographic makeup of the target audience you are trying to influence.
Ash Lockett:
I couldn't agree more. It's also about ensuring you're surveying the people who actually feel the pain point. If you want to talk about the practical pitfalls of software deployment, don't just survey the C-suite who bought the tool, survey the systems administrators and engineers who are actually managing it within their daily workflows. Their insights will be significantly more granular, authentic, and media-friendly.
What are the most common pitfalls you see corporate teams fall into that cause a research project to stall or fail to deliver a return on investment?
Andrew White:
Rushing the questionnaire design phase is the biggest mistake. Teams are often under immense pressure to get a survey into the market quickly, which leads to lazy, self-serving questions. If your survey yields data where the audience reads the headline and thinks, "Well, of course they'd say that, the survey was rigged to prove their point," you have completely wasted your investment. It can also actively damages your brand's credibility.
A great questionnaire must leave room for the unknown. It is perfectly fine to have baseline data that validates an industry hunch, but you must bake in questions where you genuinely don't know what the outcome will be. That's where you discover the surprising, disruptive data points that drive headlines and command attention.
Ash Lockett:
We call those the "joker cards". When we design research briefs for Babel's clients, we ensure the core questions cleanly validate the primary campaign narrative, but we always throw in one or two high-risk "joker" questions, where the outcome is entirely unpredictable. If they land flat, it doesn't break the campaign infrastructure, but if they hit, they unlock phenomenal, counter-intuitive insights that can elevate the entire report. Taking that brave step is often how you secure your best data.
Ash Lockett:
As we wrap up, (you know me) I want to ensure our listeners walk away with immediate, practical value. If a B2B marketing leader is sitting at their desk planning a research campaign for the coming months, what are your top three tips for success?
Andrew White:
First, do your homework. Deeply analyse your end audience before writing a single question, ensuring you map out what they genuinely need to learn, not just what you want to tell them.
Second, be bold with your methodology. Don't just default to a standard, basic questionnaire. Explore advanced techniques, like building maturity models that allow prospects to score their own business health, or weave qualitative, deep-dive interviews alongside your quantitative stats to give the report a human voice.
Third, look for monetary data hooks. Wherever possible, design your questions so you can extrapolate the percentage results against broader financial or economic data. Shifting a data point from "40% of firms experience downtime" to "IT downtime is costing UK mid-market firms £4.2m annually" completely changes its impact. It instantly commands boardroom attention. Ultimately, aim to make your research educational, not just interesting.
Ash Lockett:
Brilliant advice. The moment you can tie an industry challenge to a hard pound or dollar figure, the commercial impact of your content skyrockets.
Andrew, thank you so much for your time today, and for partnering with us to bring the Mental Av-AI-lability Index to life. We'll drop the links to the full report and Sapio Research in the show notes below.
To close things out, we ask all our guests: what is the bravest piece of research or analysis you've executed recently?
Andrew White:
We recently applied an advanced statistical analysis technique called multiple regression to an international study for an HR tech brand. It essentially maps out the hidden drivers behind complex workforce issues, isolating exactly which variables cause employee churn. We extrapolated that data with financial metrics to show operational leaders the precise monetary savings they make by adjusting specific workplace policies. It was a massive mathematical lift, but creating a highly educational tool that directly impacts corporate bottom lines was incredibly rewarding.
Ash Lockett:
That sounds fascinating! I'm definitely going to have to corner you for a coffee to dig into the maths on that one. Andrew, thank you again for joining us, and thank you to everyone for listening. Check out the show notes to download the index, and we'll see you all next time on