Consumers are asking AI
Editor’s note: Jon Arthurs is a managing director at Toluna and leads the Global Sustainability Practice at the Toluna Group. With over 30 years of industry experience gained in both client-side and consulting roles, Arthurs is passionate about developing products, brands and communications for the good of people, planet and profit. Find Arthurs on LinkedIn.
Artificial intelligence is rapidly changing how consumers discover, compare and evaluate brands. The shift is already measurable: Almost half of consumers now turn to AI for help during buying journeys, while the share replacing traditional search engines with generative AI tools for product or service recommendations has more than doubled, from 25% in 2023 to 58% in 2025 (Capgemini Research Institute, 2025).
For insight teams, the largest shift we see if that AI is beginning to act as an intermediary in the consumer decision journey. Consumers are increasingly using AI tools to understand categories, compare options, summarize reviews, reduce risk and create consideration sets. Sometimes this happens through general-purpose AI assistants. In other cases, it happens through AI summaries in search, retailer assistants, marketplace tools, brand-owned chatbots, review summaries, recommendation engines or emerging personal agents.
This creates an interesting new challenge for brand research: the need to understand what AI systems surface, frame and recommend.
From category entry points to prompt entry points
Brand tracking has traditionally measured brand salience, distinctiveness, emotionally resonant and easy to find and buy. Some also explore category entry points: the needs, occasions and decision contexts that trigger category consideration. In an AI-mediated world, those category entry points increasingly become prompt entry points.
A consumer might not simply think “detergent.” They increasingly ask an AI tool: “What is the best detergent for tough stains?” Or: “Which laundry brand is best for sensitive skin?” The AI system then translates that need into a short-list, explanation or recommendation, meaning insight teams need to ask new questions:
- For which prompts does the brand appear?
- For which needs or occasions is the brand associated?
- Which competitors are mentioned alongside it? How do they compare?
- What language does the AI use to describe the brand?
- Which sources appear to shape the answer?
- Does AI frame the brand in a way that supports or weakens its intended positioning?
This is an additional layer to conventional brand tracking. Essentially, the unit of analysis expands from the survey association to the prompt-response environment.
Why this is not just an SEO problem
It’s tempting to treat AI visibility as an extension of search optimization but that would be too narrow.
AI systems retrieve brand websites and draw on a broader dataset: FAQs, product pages, expert reviews, comparison sites, marketplace listings, customer reviews, media coverage, forums, social content, structured data and operational signals such as availability, complaints and service experience.
AI-mediated brand perceptions are likely shaped by three layers of evidence:
- Owned evidence: This includes the brand’s website, FAQs, product pages, help centers, campaign pages and transcripts. This layer tells AI systems what the brand says about itself.
- Earned evidence: This includes news coverage, reviews, awards, expert commentary, comparison sites and customer forums. This layer tells AI systems whether others appear to agree.
- Operational evidence: This includes service quality, pricing, availability, complaints, delivery, returns, loyalty and issue resolution. This layer tells AI systems whether the brand experience supports the brand promise.
For insight teams, this creates a more complex research problem. A brand may be very familiar to consumers but poorly represented in AI-mediated environments if the evidence around it is unclear, fragmented or difficult for AI systems to interpret.
Emotion still matters, but it becomes evidence
One potential concern of AI’s impact is that emotional brand-building becomes less important because AI systems process information in a more functional way. That interpretation is wrong. Emotion still matters because people still feel, remember, prefer and choose. Even though AI systems do not of course experience emotion directly, they are able to infer it from the evidence available to them such as reviews, sentiment, repeated descriptions, testimonials, social proof, complaint patterns, loyalty signals and third-party commentary. As an example, Reddit shows us why emotion still matters in AI-mediated brand choice because it’s not just a forum; it’s a large, machine-readable archive of consumer sentiment, experience and peer validation – exactly the kind of evidence AI systems can use to infer how people really talk about, trust and compare brands.
A brand’s positioning may want to be perceived as quality, premium, joyful, innovative or trustworthy. If those associations are not visible, AI may reduce the brand to functional claims such as price, features or availability. This is especially important for creative assets, where brands express emotion through video, animation, music, visual metaphor, tone and atmosphere. These assets may work powerfully for humans but remain unreadable to AI unless they are supported by transcripts, metadata, descriptions, captions or explanatory copy. A practical guideline for marketers and researchers is to build emotion for people, but make sure the emotional meaning leaves a machine-readable trace.
What should insight teams measure?
The end game is to determine a validated and predictive AI availability metric. The first step is to add a small number of practical diagnostics to existing brand research. A useful starting point is to ask whether consumers are encountering AI in the category at all and not limit it to AI search. Relevant touchpoints may include general AI assistants, AI-generated search summaries, retailer or marketplace specific assistants, brand website chatbots, AI-generated review summaries, product recommendation tools, voice assistants or AI tools used by sales and service teams.
The next question is whether those tools affected consideration, making it a simple question that helps distinguish between three outcomes:
- AI made the brand more available.
- AI had no effect.
- AI made the brand less likely to be considered.
For more depth, we can then explore the role AI played in the journey. Did it introduce new brands? Compare alternatives? Summarize reviews? Help judge trust or risk? Create a consideration set? Reinforce an existing choice? Steer the consumer toward a competitor?
Alongside survey measurement, brands should consider structured AI-output audits. These should use prompt entry points derived from real consumer needs, not generic prompts, with the objective to understand whether the brand appears, and its prominence; whether it is recommended or just mentioned; whether the framing is positive or negative; and which competitors are surfaced instead.
What should CMOs and insight leaders do next?
The rise of AI-mediated brand choice creates a shared challenge for both marketing and insight leaders. It’s not just a simple technology issue, and it should not sit only with SEO, digital or e-commerce teams. If AI systems are influencing which brands are surfaced, compared and recommended, then AI availability becomes a brand strategy and equity issue.
For CMOs and insight leaders, the implications are that:
- Brand equity becomes a discovery asset, not just a persuasion asset. Historically, brand equity helped brands win attention, memory and preference among consumers. In AI-mediated journeys, it may also affect whether a brand is named, compared, cited or recommended by an AI system. Meaning that long-term brand-building should be defended not only as an upper-funnel investment, but as part of the infrastructure that helps brands remain available in machine-mediated decision environments.
- CMOs should treat AI availability as a new brand management discipline. Existing brand KPIs do not need to be replaced. Teams should add a diagnostic layer that asks whether the brand is visible, trusted, accurately represented and easy for AI systems to recommend. Then they can create action plans on how to address areas of AI availability weakness or capitalize further on a strength.
- Brand content needs to combine emotional resonance with machine-readable evidence. Brands still need distinctive, emotional work for consumers. However, they also need structured evidence for machines: strong FAQs, product claims, comparison pages, transcripts, metadata, review strategies, expert validation and clear proof points.
- Reputation and earned media become algorithmic inputs into brand equity. Reviews, expert commentary, comparison sites, forums, media coverage and operational experience can all shape how an AI system describes the brand. PR, customer experience, review management, content, search, brand tracking and service quality need to be treated as connected parts of the same system.
- CMOs should separate organic recommendation from citation and paid visibility. Being recommended by an AI answer is different from being cited as a source. Being advertised near or within an AI-mediated journey is different again. These forms of visibility should not be collapsed into one measure.
A new layer of brand intelligence
AI will not replace human decision-making completely at this stage, nor is it that every brand choice will be delegated to an agent … yet. The more immediate issue is that AI is starting to shape the information environment around choice. If AI systems influence which brands enter the consideration set, how those brands are described and which sources are used to justify them, then this becomes a brand equity question. It is no longer only a technical issue for SEO teams.
Brand tracking has always adapted as consumer decision-making has changed. The rise of AI is another such moment. The next generation of brand research will need to measure not just what people remember, but also what machines retrieve, repeat and recommend.