Methodology Disclosure

Scope: Architectural Version: v1.0 Last Updated: May 2026

This page documents the architectural shape of the Motionworks measurement system. It is published for transparency at a level appropriate for sophisticated integrators and end users evaluating the Motionworks Services.

This is an architectural disclosure, not a methodology specification. Specific model parameters, weights, exact panel composition, partner identities, and operational tunings remain proprietary trade secrets of Motionworks AI, Inc.

The Methodology Disclosure does not authorize reverse engineering, reconstitution, or development of competing measurement products. See the Acceptable Use Policy for the full set of restrictions.

1. Panel construction principles

The Motionworks panel supports population projection at national scope across the U.S. out-of-home advertising market.

Architectural principles

Probability-based recruitment frame anchored on U.S. Census-aligned demographic strata.
Continuous panel refresh on a defined cadence to maintain representativeness over time.
Movement-pattern observation derived from consenting panelists under privacy-preserving collection practices.
Calibration weighting applied at projection time to align observed panel composition to target population strata.

Not disclosed. Panel size, exact strata weights, partner-provider identities, recruitment incentive structures, retention controls, and operational tuning of the calibration model.

2. Model versioning and refresh cadence

Motionworks measurement outputs are produced from a versioned model that is updated on a defined cadence.

Architectural framework

Major model versions are released on a documented schedule with advance notice to paid-plan customers. Major versions may include changes to panel composition methodology, calibration approach, or identity-graph construction logic.
Minor model updates are released as needed for accuracy and integrity, including market expansions, panel refresh integrations, and source data updates.
Notification: paid-plan customers receive advance notification of material model version changes that affect derivative outputs.

Not disclosed. Exact release dates, specific version numbers, internal change logs, and the full set of inputs to version-promotion decisions.

3. Identity-graph framework

The Motionworks identity graph supports cross-format reach and frequency, audience overlay matching, and derivative outputs at population scope.

Architectural principles

Privacy-preserving identity resolution using deterministic and probabilistic signal types.
Layered architecture separating raw signal ingestion, identity resolution, and downstream output construction.
Strict separation between panelist-level observation and population-level projection. Population outputs do not expose panelist-level identifiers.
No re-identification of individuals from Motionworks outputs is permitted under any plan.

Not disclosed. Exact signal sources, weighting between deterministic and probabilistic resolution, and the proprietary logic that drives identity-graph construction.

4. Source categories

The following source categories contribute to Motionworks outputs.

Disclosed at category level

Probability-based panel observation.
Census-aligned demographic reference data.
Out-of-home inventory reference data, including format, location, and operational metadata sourced from operators and validated by Motionworks.
Mobility and movement-pattern signals from panel-aligned sources under privacy-preserving collection.
Cross-format exposure signals where contractually available.

Not disclosed. Named partners, contract terms, partner-level data volumes, and specific source weighting in the model.

5. Calibration and projection

Population-level outputs are projected from observed panel data using calibration weighting tuned to target population strata.

Architectural posture

Calibration is applied at projection time, not at panel construction time.
Multiple calibration paths are evaluated against held-out validation cohorts.
Outputs include confidence framing where applicable to the metric class.
Material calibration changes are released as major model version updates.

Not disclosed. The specific calibration loss function, validation cohort composition, hyperparameters, and held-out performance metrics.

6. Engagement under NDA

Customers on the Strategic plan receive deeper methodology engagement under NDA as part of the Strategic Order Form. This engagement may include:

Panel-construction architectural review (without partner identity disclosure)
Model version change-log review with advance notice
Calibration architecture briefing
Validation cohort design briefing

Customers on the Enterprise plan may request a one-time methodology architectural briefing as part of onboarding, subject to NDA.

7. What this disclosure does not authorize

This disclosure is provided for transparency. It does not authorize any party to:

Reverse engineer the methodology from disclosed architectural principles.
Reconstitute Motionworks outputs from third-party data using disclosed framework descriptions.
Develop, fund, market, or operate any product that is held out as an alternative to or replacement for Motionworks measurement.
Make claims about Motionworks methodology that go beyond what is disclosed on this page.

The Motionworks measurement standard remains the proprietary intellectual property of Motionworks AI, Inc. Architectural disclosure does not constitute a license to the methodology itself.