AI-powered search has moved from experiment to operating norm in UK B2B buying. A brand new study [email wall] of 175 UK business decision-makers, conducted by Norstat for Clarity Global, finds that AI tools now shape discovery, evaluation, and justification in most B2B purchase processes, with implications for marketing spend, channel mix, and internal coordination.
The most consequential finding is the depth of adoption. 79% of pros report using AI every day or weekly of their work, with one third using it day-after-day. 64% spend one to 4 hours each week using AI to make business decisions, and 80% devote no less than an hour weekly to AI-ordained decision-making. The paper states AI has turn out to be embedded within the routine workflows of buyers and brand selection.
The AI chooses
Between 52% and 59% of B2B buyers now rely more on AI summaries, use traditional search less, visit fewer web sites, read fewer long articles, and spend less time understanding raw information. 59% say they spend less time gathering knowledge and more time on assessing what an LLM produces for them. The report describes this as a compression of discovery: Buyers encounter fewer primary sources and more brokered, subjective summaries. The window during which a brand can influence perception has narrowed, subsequently, as perceptions of a brand are filtered through an LLM lens.
At the highest of the funnel, 87% of buyers use AI outlines to have an algorithm dictate what they need to read. During shortlisting of potential decisions, 65% depend on AI for vendor selection. In evaluation, 77% substitute AI for due diligence and technical assessment. 75% use AI to create or influence internal business case development. The pattern is consistent. AI doesn’t sit on the margins of research, but creates reading lists, filters suppliers, makes technical comparisons, and forms internal justification.
What AI means for brand marketers
For marketing leaders, the operational implication is obvious. Influence increasingly is dependent upon how AI systems interpret and summarise a brand. The report argues that content not structured for AI summarisation risks exclusion from consideration by buyers. In practical terms, this shifts emphasis from volume of content to the sensible and technical presentation of messages. Claims need to be direct, and messaging needs to withstand software interpretation and parsing.
The document introduces the concept of Generative Engine Optimisation, or GEO, and frames it as less mature and unpredictable compared to traditional search engine optimization practices. The report characterises AI search as a “black box”, noting that model updates, training data, and answer logic remain opaque, as stays the case when reverse-engineering traditional search engine algorithms. The paper cautions against overreacting to trends, and to treat emerging practices as hypotheses. This reflects uncertainty because the evidence base for specific optimisation methods remains to be patchy.
The reliable third-party
Channel strategy emerges as a second effect. AI models draw on an unknown range of signals, combining owned content, search visibility, and social presence with third party validation from non-objective sources corresponding to PR and commercially-driven market evaluation. Integration in channels matters because AI systems aggregate claims. The report states that at the highest of the funnel, AI answers draw less from owned content and more from third-party sources. At the underside of the funnel, when buyers request recommendations like “best” products, algorithms prioritise information corroborated on the net, with third party sources, prioritised by opaque mechanisms, treated as more less subjective than content produced by vendors.
The spending implication is a change in reallocation. Technical search engine optimization stays relevant, including site performance and structured data, since it’s assumed that AI models extract and interpret web content in the identical way that traditional search algorithms have previously. Content strategy stays central, particularly material that features named experts, data referenced from third-parties, and impartial customer reviews. PR and analyst relations gain weight because they’re considered authoritative. The report advises prioritising placements over link volume and specializing in quality, relevance, and fresh citations. For budget holders, this points to sustained investment in earned media and positioning by those deemed experts.
Measurement and guesswork
Measurement is presented as vital, yet problematic. Standard analytics show activity on owned channels but don’t reveal how a brand appears in AI-generated answers. AI responses are described as “wholly ephemeral,” as these tend to vary by prompt, model, time, and context. The report recommends corporations “”look for a tool that uses automation to simplify AI search monitoring, together with real-time metrics” (but doesn’t cite a particular solution), and to track results over time to assess whether AI reproduces an organisation’s claims, quite than simply mentioning the brand.
A final operational theme from the report is internal alignment, arguing that inconsistent terminology in website, spokespeople, and sales materials can create mixed signals for AI systems, and by proxy, potential buyers. It recommends a process during which messaging is clarified, integrated into channels, measured in AI search outputs, after which refined monthly or quarterly.
The study is proscribed to 175 UK decision-makers, but provides evidence that AI is now an element of the B2B buying journey. For marketing decision-makers, the load of evidence supports three priorities: treat AI search as a primary discovery channel, align messaging and proof to withstand the vagaries of buyers counting on LLMs to make their decisions, and re-allocate spend towards integrated content and third-party validation.
(Image source: “Supermarket shelves” by Frankie Fouganthin is licensed under CC BY-SA 4.0. )
Find out more concerning the Digital Marketing World Forum series and register here.
Read the total article here
