Brand measurement has leaned on the wrong crutch for decades. Surveys and panel-based metrics from firms like Kantar, Ipsos and NielsenIQ turned something very hard to observe into something structured, but they capture only part of what happens between a consumer’s experience and the final purchase. The argument now emerging is that AI makes a multisource view practical and affordable.
The gap between what people say and what they do
Traditional brand measurement asks panels of humans what they think, then converts those answers into awareness, equity and valuation. The weakness is not the modelling. It is the input: people are unreliable narrators of their own behaviour. A shopper may say he prefers local businesses while his order history points elsewhere. A traveller may say service and comfort drive airline choice, then book on price.
Between a consumer’s experience with a brand and a survey checkbox, there are several layers of information loss. The contributor maps them: marketing happens, consumers experience it, perceptions change or don’t, researchers ask, consumers try to remember and verbalise, and a model converts the answers into metrics.
What an augmented model measures
Instead of replacing surveys outright, the piece argues for always-on measurement validated by observable outcomes. The proposed model uses three pillars:
- Mental availability: search volume, traffic and AI prompt analysis that show whether a brand is top of mind during research.
- Perception: social and ecommerce interactions that are evaluated, contextualised and streamed continuously.
- Commercial power: owned and retailer price elasticity, plus stock price as a signal of durability.
AI discovery adds a new observable surface. A consumer prompt may not name a brand, but the AI-generated response reveals what the model considers. That agentic brand perception will shape what users see and eventually what an AI buys.
Why this changes budget decisions
The article shares a client example: a high-dollar branding campaign launched against internal headwinds. A concurrent brand lift study would take about a month to show results, but the team could monitor share of search and perception in near real time. That made the budget defensible earlier and let investment decisions happen while there was still time to affect the outcome.
The practical rule: index and weight
All signals in this approach are indexed against competitors, so a positive shift is attributable to the brand and not to a category trend or seasonality. The pillars are not equal: each one is statistically weighted by its ability to predict revenue outcomes. This turns brand measurement from an abstract score into a budgeting input.
What marketers should do now
Start by identifying the observable proxies you already own: share of search, social sentiment, price elasticity and AI answer presence. Build a lightweight dashboard before commissioning the next big survey. Then use surveys as one validation input, not as the source of truth.
Source: MarTech
