The rise of generative video tools has solved a long-standing bottleneck in advertising: producing creative assets at scale. What used to take days or weeks can now be rendered in minutes, but that headlong surge in supply has created a new problem for marketers—an acute decision bottleneck. Brands can generate dozens of candidate ads per day, but choosing which ones to fund with media spend is becoming slower, pricier and more error-prone than ever.
Chinese adtech firm Minglue Technology (listed 2718.HK) has launched a product called AdEff that aims to shift the AI-in-marketing wave from the production side to the decision side. The tool combines multimodal AI with neuroscience measurements—chiefly eye-tracking and electroencephalogram (EEG) data—to evaluate video creatives in a matter of minutes and flag the highest-potential variants before media budgets are committed.
Minglue says AdEff can complete an ad test in as little as 15 minutes, modelling audience attention curves, emotional fluctuations and other cognitive signals to identify so-called ‘golden’ moments within a spot—early seconds that hook viewers, brand reveal moments and interest peaks. The system rests on the company’s self-developed hypergraph multimodal large language model (HMLLM), two original datasets (Video-SME and SPA-ADV) built from more than 100,000 neuroscience measurements, and a mixture-of-experts architecture that dispatches specialised models to interpret different signal types.
The company presents two quantitative validations: an 89% correlation between AdEff’s predicted scores and scores from human test samples, and a 76% concordance rate between AdEff’s composite assessments and industry experts’ expectations across multiple evaluation metrics. These figures underpin Minglue’s pitch to fast-moving consumer goods, healthcare, beauty, consumer electronics and e-commerce advertisers, and to independent creators who now work at the intersection of production and media buying.
If accurate and robust, the proposition matters. For marketers drowning in AIGC-produced permutations, rapid pre-flight screening can reduce wasted media spend, shorten campaign cycles, and sharpen creative iteration. The technology also promises to professionalise what has otherwise been a subjective process: screening by gut, small-scale pilot buys that exhaust budgets and miss timely windows, or slow, costly lab tests.
But the approach raises questions. Biometric proxies for persuasion—attention spikes, EEG signatures—are imperfect measures of long-term brand effects, purchase intent and social amplification. The datasets and models supporting AdEff were developed by Minglue and while parts are open-sourced, independent third-party validation, cross-cultural robustness and the representativeness of subject samples will determine how reliably physiological signals predict market outcomes. Privacy and consent around collecting and using biometric data are other practical and regulatory constraints that advertisers must navigate.
Minglue’s move also signals an industry inflection: adtech is moving beyond automated creative production into automated creative curation and decisioning. Integration with AIGC pipelines—where generation, testing and optimisation operate in a continuous loop—could compress campaign lead times dramatically. That makes AdEff a potentially valuable piece of infrastructure for marketers seeking to convert the promise of rapid creative production into reliable return on ad spend.
For global readers, the story is also about competitive dynamics. A Chinese-listed firm touting neuroscience-backed ad evaluation joins a crowded international field of attention- and emotion-measurement providers. Whether Minglue can scale its science, meet privacy standards across markets, and translate surrogate metrics into consistent commercial uplift will determine if AdEff is a niche tool or the start of a new class of decisioning platforms for AI-era marketing.