Synthetic audiences are everywhere now, but the question that actually decides whether one is worth trusting is rarely asked: how much of a real person does the model need before it behaves like the original?

In a Columbia study, more than two thousand people each answered five hundred questions about themselves and an AI twin was built of every one, and the twins reproduced people’s held-out answers at about eighty-eight percent of the rate the humans matched their own answers two weeks later.

The striking part: more than a dozen ways of building the twin — different models, formats, prompting tricks, even fine-tuning — barely changed the result. What carried it was the depth of the data, not the cleverness of the model.

The same lesson shows up when Bain backtested synthetic audiences against a company’s real prior study and concluded the data grounding a model matters more than the model itself, and when BCG reported a synthetic panel predicting real shoppers’ choices with ninety-two percent accuracy — after fine-tuning on real data. A working paper from University College Dublin adds the other half: demographics alone can’t capture how a real person decides, and the data has to be organized for the job it’s doing.

The takeaway for brand leaders: when someone hands you a customer model, stop asking which AI is under the hood and start asking how much real, first-party data it’s built on — and whether it was tuned for the decision you’re making.

Large language models are now embedded in how many brands do market research — simulate consumer responses, get results in hours instead of weeks, at a fraction of the cost.

But a rigorous Harvard Business School and Microsoft Research working paper finds that LLM-based preference estimates are realistic in some contexts and wrong-signed in others — with estimates sometimes off by a factor of three — and that a researcher without a human benchmark has no basis for knowing which is which. A Journal of Marketing study finds the human-LLM hybrid outperforms both approaches alone, but only when the human judgment is genuine rather than a pass-through. And research published in Harvard Business Review finds that LLMs consistently recommend strategies aligned with managerial buzzwords rather than context-specific logic — a pattern the researchers named trendslop.

Taken together, the evidence points to a specific map: LLMs for market research are reliable supplements when you have prior human data from the same category, unreliable substitutes when you don’t, and systematically biased toward the fashionable on exactly the questions where conventional industry thinking most needs challenging.

This episode makes that map practical — three questions a brand leader should ask before acting on what AI-assisted research tells them.

Synthetic respondents have gone mainstream — AI stand-ins for your customers you can question without recruiting anyone, a focus group that never closes.

But not all of them are worth listening to, and this episode is about what actually decides that: not the fact that they never end, but the depth and realness of the data the model is built on, and whether it was built for the job you’re asking it to do.

We draw on a Stanford study where AI versions of real people, built from rich interviews, far outperformed versions built from demographics alone — and on its 2026 follow-up breaking the result down by how much real data each model was given — alongside new work finding that piling demographic detail onto a generic model doesn’t reliably make it more accurate, to get specific about what these modelled respondents are, why a stand-in is only ever as good as the data underneath it, and the two questions a brand leader should ask before trusting one — what is it built on, and what was it built for?

A model grounded in deep, real customer data is an instrument; a thin guess in a persuasive font is a mirror.

For decades, brands have understood their customers through age, gender, and income — but those categories are losing their power to predict what people actually do.

This episode of Know Your Audience digs into Adobe’s 2026 consumer research, where a survey of four thousand people found that attitudes toward AI tracked comfort with the technology far more than any demographic line, and into recaps from this year’s IIEX APAC sessions showing how the rise of one-person households is quietly rewriting assumptions about who the customer is — alongside a reminder that behavioral data only means something in cultural context.

It's a shift even the most careful institutions are acknowledging — Pew Research Center now treats generational labels with caution, and McKinsey finds the segmentation that works is built on attitudes rather than demographics. The takeaway for brand leaders: relevance is now earned moment to moment, not assigned by segment, and the teams reading their customers a layer deeper are the ones changing the first question they ask.

Two findings from the same week tell one story. Consumers say seven in ten AI-made ads are missing their soul, while marketers are being told the smartest investment isn't creative at all but the metadata that lets AI systems understand a brand. Put together, they reveal that brands now serve two audiences with opposite taste: the human who wants soul, and the machine that decides whether the human ever sees them. In agentic commerce, an agent can rule a brand out for thin data before a customer appears; in AI search, audience preference is becoming machine visibility; and the human is pushing back, with a surge toward AI-free search. Underneath the split, both audiences chase one currency - trust - which is why SEO thinkers now argue that, as AI summarizes the web, brand recognition outweighs ranking. The takeaway for brand leaders: the job has doubled - be legible enough to be found by the machine, and human enough to be chosen by the person - and the next scorecard worth building is how the machines perceive you, right next to how humans do.

Braze’s twenty twenty-six Global Customer Engagement Review found that ninety-three percent of marketing leaders believe AI helps them accurately understand customer needs — but only fifty-three percent of consumers agree brands are successfully predicting their wants.

That forty-point confidence gap is this episode’s starting point. New academic research from Carnegie Mellon and Microsoft suggests that frequent AI use is associated with measurable declines in critical thinking, and a study of nearly two thousand workers found that only three in ten say their workplace fosters curiosity — a finding that Forbes explored this week in the context of AI’s hidden cost to original thinking at work.

Meanwhile, the evidence of what’s possible when customer intelligence connects directly to business outcomes is real and specific: a major restaurant chain attributed a thirty percent increase in recommendation likelihood to AI-powered research that reshaped its menu decisions, and a health system built predictive models catching missed appointments and prescription delays before they became gaps in care. But in the same week, a major analyst firm assessed one of the largest acquisitions in the CX industry and concluded a significant platform is unlikely to survive as a standalone product, advising its clients to start preparing for a competitive evaluation. And on retail shelves, emotion recognition technology is reading shoppers’ facial expressions in real time and adapting content accordingly — while the peer-reviewed science behind facial-expression-to-emotion mapping remains contested and the EU classifies the technology as high-risk.

This episode unpacks why the gap between how well brands think they understand their customers and how well that understanding actually works may be widening — and what a CMO should be pressure-testing right now.

Know Your Audience is a weekly podcast about how the nature of understanding your customer is changing — and what that means for the marketers making decisions because of it.

Produced by Soulmates AI.

The fastest-growing AI habit among consumers isn’t creating content or learning new skills — it’s shopping. Forty-two percent of consumers used AI to shop in the past month. Product discovery is the single highest-adoption personal AI task tracked across fifty-four categories, and it’s the only one that’s both widespread and still growing. Your consumer is comparing options, interpreting reviews, and narrowing choices inside AI conversations before they ever reach your website, your app, or your store. But most consumer brands measure discovery through search rankings, paid media impressions, and brand tracker scores — surfaces that assume the journey starts on a search engine or a social feed. The AI-mediated portion of the shopping journey is largely invisible to current measurement infrastructure. The gap between where the customer actually starts shopping and where the brand thinks they start is widening every quarter.

Show Notes:

42% of Consumers Now Use AI Tools to Shop, NIQ Data Shows

NIQ research showing AI is becoming embedded in how consumers shop, influencing evaluation, comparison, pricing, and choice narrowing. Based on monthly survey of approximately five hundred consumers across the United States.

NielsenIQ, May 5, 2026

https://nielseniq.com/global/en/news-center/2026/42-of-consumers-now-use-ai-tools-to-shop-niq-data-shows/

Consumers Are Learning to Trust AI Through Shopping

PYMNTS Agentic AI Report Series tracking fifty-four personal AI tasks over five months. Product discovery identified as the highest-adoption, most stable AI habit forming among consumers, with universal reach across age, income, and gender.

PYMNTS Intelligence, May 14, 2026

https://www.pymnts.com/artificial-intelligence-2/2026/consumers-are-learning-to-trust-ai-through-shopping/

IBM-NRF Study: Brands and Retailers Navigate a New Reality as AI Shapes Consumer Decisions Before Shopping Begins

Global study finding forty-five percent of consumers turn to AI during buying journeys. Forty-one percent use AI to research products, thirty-three percent to interpret reviews, thirty-one percent to find deals.

IBM Institute for Business Value / National Retail Federation, January 7, 2026

https://newsroom.ibm.com/2026-01-07-ibm-nrf-study-brands-and-retailers-navigate-a-new-reality-as-ai-shapes-consumer-decisions-before-shopping-begins

How Your Customers Use AI to Shop (2026 Trends)

Consumer survey finding forty-one percent of consumers purchased a product AI recommended in the past six months. Gen Z and millennials twenty-one percent more likely to buy based on AI suggestions.

Klaviyo, 2026

https://www.klaviyo.com/marketing-resources/ai-consumer-trends

Adobe 2026 AI and Digital Trends: Customer Behaviors and AI

Global survey of four thousand consumers. About a quarter now cite AI-powered platforms as their top research tool, ahead of brand websites and online reviews.

Adobe, March 2026

https://business.adobe.com/resources/digital-trends-consumer-report.html

AEO Software Category Grows Over 2000% on G2

G2 reporting that the answer engine optimization software category grew from seven products to over one hundred and fifty in ten months.

G2 / PR Newswire, January 30, 2026

https://www.prnewswire.com/news-releases/aeo-software-category-grows-over-2000-on-g2-as-half-of-b2b-buyers-start-their-search-with-ai-chatbots-over-google-302674557.html

The CX platforms that many brands depend on are under serious financial pressure — Medallia’s five-billion-dollar equity wipeout is the most visible example, but the structural forces behind it apply across the category. At the same time, the CMO role itself is being redefined: Forrester’s latest report argues that AI is pushing marketing leaders from campaign execution into enterprise growth orchestration, while AI-native CX startups are raising at multi-billion-dollar valuations. The old infrastructure is cracking and the role that depends on it is changing shape. Brand leaders who grasp both shifts together will make better decisions than those who see them in isolation.

Show Notes:

Medallia Survived, but the Real Story Should Worry Marketers

Detailed analysis of the Medallia restructuring, Blackstone earnings call commentary, lender consortium plans, and due diligence framework for evaluating PE-backed software vendors.

MarTech, May 11, 2026

https://martech.org/medallia-survived-but-the-real-story-should-worry-marketers/

David Sacks: PE Debt-Fueled Buyouts Are Now Software’s Third Major Exit

All-In Podcast discussion on AI deflation compressing SaaS pricing, net revenue retention declines, and the structural threat to PE software buyout models.

Benzinga / All-In Podcast, May 8, 2026

https://www.benzinga.com/news/topics/26/05/52432945/venture-capitalist-david-sacks-says-private-equitys-debt-fueled-buyouts-are-now-softwares-third-major

The AI CMO: Growth Accountability Gets Next-Level

Forrester report on how AI is reshaping CMO accountability, shifting the role from campaign execution to enterprise growth orchestration. Covers agentic marketing, AI as a new audience, and the shift from headcount-constrained to systems-scaled marketing.

Forrester, April 2026

https://www.forrester.com/blogs/the-ai-cmo-growth-accountability-gets-next-level/

AI Turns CMOs into Chief Growth Officers: Forrester

Coverage of the Forrester AI CMO report, including data on B2C executive growth expectations, B2B AI agent adoption rates, and consumer expectations around agentic marketing.

Campaign US, April 2026

https://www.campaignlive.com/article/ai-turns-cmos-chief-growth-officers-forrester/1956317

Sector Snapshot: Sales and Marketing Gets an AI Makeover

Crunchbase reporting on AI-native CX and marketing startup funding, including Sierra ($950M at $15B), Hightouch ($150M at $2.75B), and Netomi ($110M) rounds.

Crunchbase, May 8, 2026

https://news.crunchbase.com/sales-marketing/gtm-ai-startup-funding-netomi-hightouch/

SurveyMonkey Embeds Survey Creation and Analysis Inside Claude

Coverage of SurveyMonkey’s Claude connector launch, AI buildout since February 2026, and G2 Spring 2026 rankings.

CMSWire, May 7, 2026

https://www.cmswire.com/customer-experience/surveymonkey-embeds-survey-creation-and-analysis-inside-claude/