Thirty sources. Five fields. One thesis.

https://www.thinkwithgoogle.com/marketing-strategies/automation/2011-winning-zmot-ebook/

Authority

Authored by Google's then Managing Director of US Sales & Service; the foundational ebook that named the concept and seeded fifteen years of derivative research.

Key findings

Shoppers consulted an average of 10.4 information sources before deciding in 2011, up from 5.3 the previous year. Eighty-four per cent said the new mental model shaped their decisions.

Direct quote (11 words)

“The Internet has changed how we decide what to buy.”

Why it supports Audience Intel

Establishes the historical anchor: the decisive moment moved from the shelf to the screen a decade and a half ago. Everything Audience Intel measures sits inside this moment.

“At Google, we call this online decision-making moment the Zero Moment of Truth.”

https://hbr.org/2022/01/sensemaking-for-sales

Authority

Adamson co-authored The Challenger Sale; led Gartner's sales research practice for nearly two decades. State the fieldwork year, not the article year: the 2022 HBR piece is a vehicle for a 2017 Gartner measurement (n = 750), published in Win More B2B Sales Deals (Gartner, 2018). Gartner itself carried the 17% on its B2B buying-journey page in 2020 and 2022 and has not carried it in any capture from January 2024 onward — retired without retraction or explanation.

Key findings

B2B buyers spend 17% of the total purchase journey meeting potential suppliers. That 17% is divided across every supplier considered, leaving any single firm roughly 5–6%. The same chart splits buying-group time as: researching independently online 27%, meeting the buying group 22%, researching independently offline 18%, meeting potential suppliers 17%, other 16% — so roughly 45% of buying time is independent research.

Direct quote (14 words)

“The single biggest challenge of selling today is our customers' struggle to buy.”

Why it supports Audience Intel

A statement about seller access and share of attention: 17% is split across rival vendors, so any one firm gets 5–6%. It is NOT a statement about journey stage, and must not be used to carry a “the decision is X% complete before contact” claim — Gartner's own text rejects that framing (“B2B buying doesn't play out in any kind of predictable, linear order”). The pre-contact claim is carried by the 6sense and TrustRadius shortlist evidence, not by this figure. The ~45% independent-research split is the more directly relevant number.

“That 17% of purchase activity allocated to supplier interaction represents all suppliers, not each supplier.”

https://www.gartner.com/en/newsroom/press-releases/2026-03-09-gartner-sales-survey-finds-67-percent-of-b2b-buyers-prefer-a-rep-free-experience

Authority

Most recent Gartner sales-practice release; analyst Alyssa Cruz, Senior Principal Analyst.

Key findings

Sixty-seven per cent of B2B buyers now prefer a rep-free experience. Forty-five per cent used AI during a recent purchase. Confident buyers are twice as likely to report a high-quality deal.

Direct quote (12 words)

“B2B buyers are progressing through critical buying tasks in more autonomous ways.”

Why it supports Audience Intel

Quantifies the present-tense preference for self-service. The smaller the deal, the more absolute this becomes.

“Sellers can't rely on static collateral to carry influence in those moments.”

https://www.digitalcommerce360.com/2025/07/07/b2b-buyers-vendor-in-mind-forrester/

Authority

Forrester's buyer-side research programme. TRACEABILITY: the report is named and confirmed to exist but is client-only; the figures below are held via reputable trade press (Digital Commerce 360, 7 July 2025; Marketing Week). Say so when citing — never write “Forrester found” without noting we are citing the trade-press account of a paywalled report. Forrester does not publicly disclose the survey's sample size.

Key findings

Ninety-two per cent of B2B buyers begin their process with at least one vendor already in mind. Forty-one per cent have settled on a single preferred vendor before formal evaluation starts.

Direct quote (10 words)

“B2B buying today is a process of confirmation, not selection.”

Why it supports Audience Intel

The most direct, most recent evidence that the shortlist is formed before a supplier is contacted — the claim Audience Intel's product rests on. Replaces the retired “90% rely on generative AI to shortlist vendors” claim, which misdescribed its source (see the flagged-for-caution appendix).

“B2B buying today is a process of confirmation, not selection.”

https://www.forrester.com/blogs/b2b-marketers-its-time-to-ditch-sourcing-metrics/

Authority

Forrester's longitudinal buyer survey, tracking interactions per purchase across 2017, 2019, and 2021; n > 950 in the 2021 wave.

Key findings

B2B buyers logged 27 buying interactions per individual buyer journey in 2021, up from 17 in 2019 and roughly twice the 2015 level — 15 digital plus 12 human. Forty-three per cent of buying scenarios are now classified “highly complex.” The 27 is per journey, not per vendor: multiplying it by the number of vendors considered inflates the figure several-fold.

Direct quote (13 words)

“Buyers are demanding a fluid experience that intertwines self-service, seller-driven and marketing-assisted information flow.”

Why it supports Audience Intel

Documents the inflation of digital touchpoints — and the fact that almost two-thirds of them are unsupervised.

“The days where buyers were willing to accept salespeople controlling the flow of information are long gone.”

https://www.mckinsey.com/capabilities/growth-marketing-and-sales/our-insights/five-fundamental-truths-how-b2b-winners-keep-growing

Authority

Ninth annual edition; cumulatively over 30,000 respondents since 2016. The longest-running large-sample B2B buyer survey from a tier-one consultancy.

Key findings

B2B decision-makers now use ten interaction channels per purchase, up from five in 2016. Thirty-nine per cent are willing to spend over $500,000 through self-service or remote channels; 20% are comfortable with $1m–$10m transactions remotely. Fifty-four per cent would abandon a purchase after a single poor omnichannel experience.

Direct quote (11 words)

“Customers are willing to walk away if they don't get [it].”

Why it supports Audience Intel

Demolishes the lingering assumption that high-ticket B2B still demands human contact. If buyers will spend half a million via self-service, the sub-£2,000 decision is already decided online.

“Those in B2B sales who think investing more in e-commerce is optional are mistaken.”

https://go.trustradius.com/rs/827-FOI-687/images/2024%20B2B%20Buying%20Disconnect%20Year%20of%20the%20Brand%20Crisis.pdf

Authority

Eighth annual edition; the only large-sample study deliberately designed to surface the gap between buyer and vendor perception. Two caveats to state whenever it is used: the respondents are TECHNOLOGY buyers, not professional-services buyers, and they are drawn from a software review site's own network, which over-represents research-active buyers.

Key findings

Seventy-eight per cent of buyers shortlisted a product they had heard of before starting research, rising to 86% among enterprise buyers. Seventy-one per cent bought their top-choice product. Use of analyst reports fell to a seven-year low of 16%.

Direct quote (15 words)

“Buyers are gravitating toward familiar products they've used before or heard about from their networks.”

Why it supports Audience Intel

The shortlist is built before active research begins. By the time the website is visited, it is being checked, not chosen.

“Of buyers creating shortlists, 78% reported selecting products they'd heard of before starting research.”

https://www.trustradius.com/vendor-blog/b2b-buying-disconnect

Authority

Tenth annual edition of the same paired buyer/vendor instrument as the 2024 report. Same two caveats: technology buyers, drawn from a review site's own network. Do NOT build a trend line between this wave and the 2024 or 2025 waves — TrustRadius's own LLM-usage figure moved from 7% (Jan 2025) to 63% (Jan 2026), which is question design moving, not behaviour.

Key findings

Eighty-three per cent of buyers shortlisted three or fewer products; the average shortlist held 2.7. Sixty-three per cent used AI somewhere during the purchase journey, and 94% of those AI users fact-check what it tells them.

Why it supports Audience Intel

With 6sense's day-one shortlist finding, this is the strongest ground in the whole territory: two independent instruments, different samples, different years, converging on how small shortlists are and how early they close. If a firm is not on a list of roughly three, it is not in the running.

“83% of buyers shortlisted three or fewer products. The average shortlist was 2.7.”

https://6sense.com/resources/report/2025-b2b-buyer-experience-report/

Authority

The research programme was led by Kerry Cunningham, formerly principal analyst at Forrester / SiriusDecisions; 6sense holds large-scale B2B intent data. DISCLOSE THE INTEREST: 6sense sells intent data and has a commercial stake in the pre-contact “dark funnel” being large. The method is disclosed and reasonable, so this is a disclosure point rather than a disqualification — but name it rather than let a reader find it. Supersedes the 2024 edition (2,509 buyers), and the headline moved AGAINST the simple version of this argument.

Key findings

Ninety-five per cent of the time, the vendor that won was already on the buyer's day-one shortlist; buying groups fill roughly four shortlist slots on day one. Ninety-four per cent had ranked a preferred vendor before making first contact, and bought that favourite 77% of the time; around 80% of the time the vendor contacted first wins. First contact now falls at 61% of the way through the process, DOWN from 69% in the 2024 edition — buyers are contacting sellers earlier, and we cite 61%, not the older 69% or the free-floating “70%”.

Direct quote (7 words)

“a research assistant, not a research initiator” — 6sense analysts, on how buyers actually use LLMs

Why it supports Audience Intel

Confirms that the website is doing the qualification the salesperson used to do — silently, and without feedback. Cite the direction of travel out loud: contact is happening earlier, and the substance of the argument survives it because preference is still fixed before contact.

“95% of the time, the winning vendor was already on the buyer's day-one shortlist.”

https://www.gartner.com/en/newsroom/press-releases/2023-06-14-gartner-survey-reveals-60-percent-of-technology-buyers-involved-in-renewal-decisions-regret-nearly-every-purchase-they-make

Authority

Hank Barnes, Distinguished VP Analyst, Chief of Research for Enterprise Buying Behavior, Gartner.

Key findings

Sixty per cent of tech buyers regret nearly every renewal. High-regret deals take 7–10 months longer than no-regret deals. Eighty-nine per cent of high-regret buyers cite conflicting team objectives, versus 9% of no-regret buyers.

Direct quote (10 words)

“Regret is a known issue with enterprise technology purchase decisions.”

Why it supports Audience Intel

Buyer regret is the dark twin of supplier confidence. Audience Intel's analysis lives on the side of the screen that produces — or prevents — it.

“High-regret feelings peak before implementation — indicating frustration with the buying experience itself.”

https://doi.org/10.1080/01449290500330448

Authority

Carleton University's Human-Oriented Technology Lab; over a thousand Google Scholar citations.

Key findings

Visual appeal of websites is reliably assessed within 50 milliseconds. Ratings at 50ms correlated highly with ratings at 500ms. The follow-up Tuch et al. (2012) study at Basel and Google extended the finding to 17 milliseconds.

Direct quote (19 words)

“Visual appeal can be assessed within 50 ms, suggesting designers have about 50 ms to make a first impression.”

Why it supports Audience Intel

Establishes the temporal floor: by the time a buyer thinks, the verdict has already happened.

“You have fifty milliseconds — one-twentieth of a second — to make a first impression.”

https://doi.org/10.1111/j.1467-9280.2006.01750.x

Authority

Princeton University; the canonical study on rapid face judgement, NSF-funded.

Key findings

Hundred-millisecond exposures produced trait judgements — particularly trustworthiness — that correlated highly with unconstrained-time judgements. Longer looks did not improve judgement; they increased confidence.

Direct quote (17 words)

“Judgments made after a 100-ms exposure correlated highly with judgments made in the absence of time constraints.” — Alex Todorov

Why it supports Audience Intel

A website is a face. The judgement formed in the first tenth of a second is not corrected by subsequent reading — it is rationalised.

“It appears we are hard-wired to draw these inferences in a fast, unreflective way.”

https://doi.org/10.1037/0033-2909.111.2.256

Authority

Harvard meta-analysis of 38 thin-slicing studies; the academic foundation later popularised in Gladwell's Blink.

Key findings

Overall effect size r = .39 for the predictive accuracy of judgements based on observations under five minutes. Slice length, between 30 seconds and five minutes, did not improve accuracy.

Direct quote (18 words)

“Predictions based on observations under half a minute did not differ significantly from predictions based on five-minute observations.”

Why it supports Audience Intel

Thin slices predict outcomes. The website is the thinnest of slices, and the buyer is forming the prediction whether or not the firm is paying attention.

“Thin slices of behaviour predict outcomes as accurately as much longer observations.”

https://us.macmillan.com/books/9780374533557/thinkingfastandslow/

Authority

Nobel Laureate in Economics, 2002. Foundational dual-process account.

Key findings

System 1 (fast, automatic, intuitive) issues continuous suggestions to System 2 (slow, deliberate). Most decisions, including most professional-services hiring decisions, are System 1 verdicts ratified by System 2 rationalisation.

Direct quote (13 words)

“System 1 continuously generates suggestions for System 2: impressions, intuitions, intentions, and feelings.”

Why it supports Audience Intel

Gives a theoretical name to what the buyer is actually doing on a homepage. The decision is intuitive; the meeting is the ratification.

“System 1 runs automatically and System 2 is normally in a comfortable low-effort mode.”

https://doi.org/10.1207/s15327957pspr0804_3

Authority

Foundational paper for processing-fluency theory.

Key findings

The more easily the brain processes a stimulus, the more positively it is judged — for beauty, truth, and trust. The effect operates below conscious awareness.

Direct quote (14 words)

“The more fluently perceivers can process an object, the more positive their aesthetic response.”

Why it supports Audience Intel

A cluttered or slow website is not merely inconvenient. It is cognitively unpleasant, and that unpleasantness transfers to the firm.

“Beauty is in the perceiver's processing experience — what's easier to process feels more pleasing.”

https://doi.org/10.1037/0022-3514.79.6.995

Authority

Columbia and Stanford; field experiment at Draeger's Grocery Store.

Key findings

Three per cent of customers offered 24 jams purchased; 30% of customers offered six jams purchased — a roughly tenfold conversion gap. Subsequent meta-analyses (Chernev et al., 2010) find the effect is context-dependent.

Direct quote (15 words)

“People are more likely to purchase gourmet jams when offered a limited array of six choices.”

Why it supports Audience Intel

Professional-services sites that present every service line equally are committing the jam error.

“Six jams sold ten times more than twenty-four — too much choice paralyses buyers.”

https://doi.org/10.1145/1541948.1541999

Authority

Founder, Stanford Behavior Design Lab.

Key findings

Behaviour occurs when motivation, ability, and a prompt converge in the same moment — B = MAP. Friction reduces ability; absence of a clear prompt prevents action even when motivation is high.

Direct quote (13 words)

“Behavior happens when Motivation and Ability and Prompt converge at the same moment.”

Why it supports Audience Intel

The contact form that asks for ten fields is not a contact form. It is an ability suppressor.

“Three elements must converge at the same moment for any behaviour: motivation, ability, prompt.”

https://doi.org/10.1016/j.neuron.2004.09.019

Authority

Baylor College of Medicine; founding study of neuromarketing.

Key findings

Anonymous taste produced one neural pattern; brand-cued taste produced another. Knowing a drink was Coke altered both expressed preference and ventromedial prefrontal cortex activity, even when the liquid was identical.

Direct quote (13 words)

“Brand knowledge had a dramatic influence on expressed behavioural preferences and brain responses.”

Why it supports Audience Intel

A credible website does not merely communicate competence; it changes how subsequent meetings are perceived.

“Brand cues literally change what the brain prefers — even when the product is chemically identical.”

https://www.nngroup.com/articles/how-long-do-users-stay-on-web-pages/

Authority

Nielsen Norman Group is the principal usability-research consultancy; the underlying dataset is Microsoft Research's.

Key findings

Users leave most pages within 10–20 seconds. Page survival follows a Weibull distribution: surviving the first ten seconds raises the probability of staying minutes. The first ten seconds are decisive.

Direct quote (10 words)

“Users often leave web pages in ten to twenty seconds.”

Why it supports Audience Intel

Translates the millisecond literature into operational stakes. A site has roughly two breaths to earn the third.

“You have ten seconds to communicate your value proposition — or users vanish.”

https://blog.google/products/admanager/the-need-for-mobile-speed/

Authority

Aggregated Google Analytics data across 3,700 mobile sites.

Key findings

Fifty-three per cent of mobile visits are abandoned if a page takes longer than three seconds to load. Sites loading in five seconds had 70% longer sessions and 35% lower bounce than sites loading in nineteen seconds.

Direct quote (14 words)

“53% of visits are likely to be abandoned if pages take longer than 3 seconds.”

Why it supports Audience Intel

Speed is not a technical metric. It is the first credibility cue.

“More than half of mobile users abandon a site if it takes longer than three seconds.”

https://hingemarketing.com/library/article/inside-the-buyers-brain-fourth-edition-executive-summary

Authority

Hinge has studied over 40,000 buyers and sellers of professional services across fifteen years; the largest ongoing primary-research programme in the category.

Key findings

Buyer-perceived visibility of professional-services firms fell from 23.1% in 2020 to 14.6% in 2022 — a 37% decline. “Technology and data issues” entered buyers' top-five challenges, up 136%. Existing-relationship reliance as an evaluation method rose 79%.

Direct quote (11 words)

“Referrals are the number one way buyers discover service providers.”

Why it supports Audience Intel

Visibility — the precondition of being shortlisted at all — is collapsing precisely as buyer self-direction intensifies. Firms that cannot be found online are not chosen.

“Buyer visibility of professional-services firms fell 37% in two years — invisible firms aren't considered.”

https://hingemarketing.com/library/article/high-growth-study-2025-executive-summary

Authority

Tenth annual edition; the longest-running primary study of growth in professional services.

Key findings

High-growth firms — defined as 20%+ CAGR over three years — grow four times faster and are up to 30% more profitable. They invest roughly twice as much in marketing and prioritise educational content, differentiation, and digital maturity.

Direct quote (15 words)

“High growth firms are not just reacting to change; they are proactively shaping their future.”

Why it supports Audience Intel

Quantifies the marketing-to-growth elasticity in the category Audience Intel serves.

“The fastest-growing professional-services firms grow four times faster — and are 30% more profitable.”

https://www.edelman.com/expertise/Business-Marketing/2024-b2b-thought-leadership-report

Authority

Sixth annual edition; the premier study on the influence of thought leadership on B2B consideration.

Key findings

Seventy-three per cent of decision-makers say a firm's thought leadership is more trustworthy than its marketing. Seventy-five per cent say it has led them to research a service they were not considering. Eighty-six per cent say strong thought leadership earns an RFP invitation — but only 15% rate what they actually read as excellent.

Direct quote (15 words)

“73% of decision-makers say a company's thought leadership is more trustworthy than its marketing materials.”

Why it supports Audience Intel

The website is the cheapest thought-leadership distribution channel a small firm owns — and the one that is most often wasted.

“86% say strong thought leadership earns an RFP invitation. Only 15% rate what they read as excellent.”

https://www.edelman.com/trust/2025/trust-barometer

Authority

Twenty-fifth annual edition; the longest-running global study of institutional trust.

Key findings

Six in ten respondents now report moderate or high grievance toward business, government, and the wealthy. Sixty-eight per cent believe business leaders deliberately mislead. The UK Trust Index sits at 43 — among the five lowest of major economies.

Direct quote (13 words)

“Over the last decade, society has devolved from fears to polarisation to grievance.”

Why it supports Audience Intel

Trust is now scarce. The website is where it is either earned or forfeited before any conversation begins.

“Trust in UK institutions sits at 43 — among the lowest of major economies.”

https://www.globenewswire.com/news-release/2025/05/13/3080217/0/en/Three-quarters-of-small-law-firms-are-making-strategic-improvements-to-drive-organic-growth.html

Authority

The longest-running survey of SME law firms in England and Wales; LexisNexis is part of RELX.

Key findings

Seventy-two per cent of small UK law firms plan organic growth in 2025, up from 40% in 2023. Forty-seven per cent still have no plans to invest in technology. Eighty-one per cent of small-firm lawyers cited attracting new business as a primary challenge in 2023.

Direct quote (10 words)

“Smaller firms are prioritising control and sustainability over scale.”

Why it supports Audience Intel

The owner-managed segment intends to grow organically — which means by acquisition of clients, which means by website.

“72% of UK small law firms plan organic growth — client acquisition is the lever.”

https://www.lawgazette.co.uk/practice-management/the-sme-legal-services-market/5041605.article

Authority

The Legal Services Board is the statutory oversight regulator; Pleasence and Balmer's work is the authoritative academic record on small-business legal need.

Key findings

Roughly one-third of UK SMEs experience a legal issue, but only 8% consult a law firm. When SMEs do seek advice, only 30% choose a solicitor — the remainder consult accountants, HR advisers, or barristers. Fewer than 15% view solicitors as cost-effective.

Direct quote (12 words)

“Fewer than 15% of SMEs thought solicitors offered cost-effective legal solutions.”

Why it supports Audience Intel

Professional-services firms compete not only with one another but with the option of doing nothing, or asking the accountant. The website must answer that question first.

“Only 8% of UK SMEs with a legal issue actually consult a law firm.”

https://arxiv.org/abs/2411.10109

Authority

Stanford HAI / Park-Bernstein team. Builds on Park et al. (2023), “Generative Agents,” the foundational paper for synthetic personas. State 2024 for the paper — the Stanford HAI brief is a 2025 summary of a 2024 submission, and citing only the brief makes the work look a year newer than it is.

Key findings

LLM agents conditioned on two-hour interviews with 1,052 real participants predicted those participants' held-out General Social Survey responses at 83% (interview-only), 82% (survey-only) and 86% (combined) of the participants' own two-week test–retest consistency — against a demographics-only baseline of 74%. So the honest figure is a range of 82–86%, and the margin over a naive demographic model is roughly 9–12 points, not a chasm.

Direct quote (19 words)

“Agents replicate participants' responses on the General Social Survey 85% as accurately as participants replicate their own answers.”

Why it supports Audience Intel

The academic warrant for simulating individuals rather than guessing from demographics. SCOPE LIMIT, stated every time it is used: the agents replicated attitudes, personality inventories and economic-game behaviour — NOT vendor selection or website judgement. This paper licenses “simulating individuals is measurably better than guessing from demographics”. It does not license “synthetic personas predict how a buyer judges your website”.

“Generative agents matched real people at 82–86% of those people's own retest consistency — against a 74% demographic baseline.”

https://www.anthropic.com/news/3-5-models-and-computer-use

Authority

Anthropic primary release; the first frontier model to operate a graphical interface like a human.

Key findings

Claude 3.5 Sonnet became the first frontier model to view a screen, move a cursor, click, and type. On OSWorld, the leading agent benchmark, Claude scored 14.9% in October 2024, against ~70% for humans; by early 2026, successor models had reached the high-60s.

Direct quote (15 words)

“Claude 3.5 Sonnet is the first frontier AI model to offer computer use in public beta.”

Why it supports Audience Intel

The capability Audience Intel sells — view a website as a human would — became operationally feasible in late 2024. Before that, it was hand-waving.

“The first frontier AI model to offer computer use.”

https://hai.stanford.edu/ai-index/2025-ai-index-report

Authority

Stanford Institute for Human-Centered AI; the most-cited annual benchmark of AI progress.

Key findings

On benchmarks introduced in 2023, performance rose 18.8, 48.9, and 67.3 percentage points within a single year on MMMU, GPQA, and SWE-bench respectively. Inference cost for a GPT-3.5-equivalent model fell roughly 280-fold from November 2022 to October 2024.

Direct quote (10 words)

“AI is a civilization-changing technology — not confined to one sector.”

Why it supports Audience Intel

Establishes that the capability curve under Audience Intel's product is steepening, not plateauing.

“Scores rose by 67.3 percentage points on SWE-bench in a single year.”

https://business.google.com/uk/think/consumer-insights/navigating-purchase-behavior-and-decision-making/

Authority

Google's flagship behavioural-science research; based on a literature review of 250-plus cognitive-bias studies and an experiment simulating 310,000 purchase scenarios.

Key findings

Buyer decisions loop through exploration and evaluation rather than progressing linearly. Six biases dominate the loops: category heuristics, power of now, social proof, scarcity, authority, and the power of free.

Direct quote (13 words)

“The way people make decisions is messy — and it's only getting messier.”

Why it supports Audience Intel

Replaces the funnel with a model that survives contact with how buyers actually behave.

“The way people make decisions is messy — and it's only getting messier.”