S12.12 · Business & Professional Services
A split market where AI commoditizes custom survey research while strengthening proprietary subscription data.
Market research and business information is two businesses wearing one label: subscription data platforms, which are IP-leveraged and increasingly AI-resistant, and custom survey research, which is people-leveraged and commoditizing quickly. Traditional survey-based market research runs roughly $76-80B (ESOMAR); the broader insights-and-analytics category, data and analytics platforms included, is estimated at $140-155B (Research World/Statista, 2024-2025). The gap between those figures is worth flagging rather than resolving to a single number — it reflects genuinely different scope definitions, not measurement error. The dynamic that decides who wins: AI models need curated, proprietary data as an input, so the same technology commoditizing the custom-research side strengthens the licensed-data side.
Growth splits along the same seam. Traditional market research is growing roughly flat to low-single-digits; subscription data and business-information providers — Gartner, IQVIA and S&P among them — are growing mid-to-high single digits. Revenue concentrates in North America and Europe, with APAC driving incremental growth. NIQ, Kantar, Ipsos, GfK, Gartner and IQVIA hold large-account share, above a fragmented boutique and regional long tail.
The recurring-revenue range here is the widest in the sector. Subscription data platforms run 85-95%+ net renewal on an IP-leveraged model; custom and ad hoc survey research is one-off and people-leveraged, priced project by project with no contracted forward visibility. What commands the premium multiple is the subscription share of a given firm's revenue mix — not brand, not headcount, not geographic footprint. The economics follow: subscription data businesses carry margins above 30% behind genuine data moats built on proprietary panels and licensed content, while custom-research margins are thinner and commoditizing as generic survey and panel capacity becomes easier to replicate.
Legacy market-research firms building proprietary data and AI products are defending, not expanding — shifting revenue mix toward the IP-leveraged side before the custom-research side erodes further, with no new client budget opened in the process.
The direct threat lands on custom survey design and analyst-written reporting: synthetic-respondent panels and AI-generated reports already substitute for lower-tier custom research work. The offsetting effect is real, not theoretical. Proprietary longitudinal panels and licensed datasets are strengthened by the same technology, because AI models require curated data as an input and cannot manufacture a multi-year panel from nothing.
The boundary opening up is between "data-as-product," which is durable and AI-resistant, and "research-as-service," which is commoditizing fast. That boundary, more than firm identity, will determine which parts of this segment retain pricing power. Custom-research commoditization is already underway; full repricing of the subscription-data AI premium sits on a 2-5 year horizon.