THE ENGINE

One engine. Every channel. Real behavior at scale.

The Behavioral Biometrics Engine classifies how people actually read, scroll, and engage with content. It works on the web, in AI responses, and in social feeds. Direct behavioral measurement, every user, every session.

THE APPROACH

Direct behavioral measurement at census scale.

The Behavioral Biometrics Engine measures cognitive engagement through signals captured in standard browsers: scroll patterns, dwell time, cursor dynamics, reading depth. Every user, every session, every page. No panels, no hardware, no sampling.

DECAY MODELS

Three models of how attention falls.

Positional decay

Where in the content does attention fall? A brand mentioned first in an AI response reaches 77% of attentive users. Mentioned last, 18%. On web pages, a brand at 50% article depth reaches 44% of readers. The same mathematical function describes both, with different calibration per channel.

Temporal freshness decay

How quickly does content lose its pull after publication? A Facebook post loses most engagement within hours. A web editorial holds for weeks. An AI response is always fresh. Each channel has its own rhythm.

Session engagement decay

How does cognitive focus change within a single reading session? Users who have been reading for 10 minutes are measurably less engaged than at minute one. BBE tracks this in real time.

THE CROSS-CHANNEL FINDING

One model. Every channel. Empirically validated.

The same decay model works across every channel BAX measures. The calibration parameter differs, the mathematical structure is identical. This is an empirical finding validated on 5M+ URLs, 10,000+ YouTube videos, and AI citation data across 300+ model variants.

Web editorial and YouTube long-form produce nearly identical decay curves. AI responses decay slightly faster. Social feeds decay 4-6x faster than editorial. One instrument reads them all.

CLASSIFICATION

Seven reader segments. Classified in 15 seconds.

Every user session is classified into one of seven cognitive types within 15 seconds of behavioral signals: Deep Reader, Active Explorer, Targeted Scanner, Flow Scroller, Headline Skimmer, Distracted Browser, Content Binger. A Deep Reader has roughly twice the brand recall probability of a Flow Scroller. BAX weights every attention metric by segment.

AUDITABILITY

Fully auditable. Open methodology.

Every BAX Index result can be independently replicated. The methodology is closed-form, the weights are explicit, and all raw data is stored in EU infrastructure. Full methodology documentation is available under NDA for enterprise clients, auditors, and regulatory bodies.