Definitions, Standards and Audits for Synthetic Data

In the US, independent certification body Synthetic Data Standard has launched its first service, a Field Guide defining six types of synthetic data used in consumer and market insights; and plans to launch a third party provider audit program called SDX-1, before the end of the year.

Synthetic Data Standard says that while the technology is moving into everyday use in the sector, shared language is lacking and providers use the same terms to describe a wide variety of offerings. The group itself defines synthetic data as 'simulated respondents, audiences and answers generated by models rather than collected from people'. SDX-1 will assess whether a provider's methods meet a defined bar for fidelity, transparency and bias handling.

The new Field Guide classifies every synthetic research solution by two questions: who does it speak for, and what data drives its answers. This leads to six types: Digital Twin, Derived Persona, Grounded Persona, Derived Panel, Grounded Panel and Sample Boost. The Guide gives a definition distinguishing each of the six types: for example, a Derived Persona portrays a typical customer from general sources such as social media, whereas a Digital Twin is modeled from one real person's own consented data and can be checked against that person. It goes on to describe the kinds of questions each type answers reliably and the kinds it does not.

The guidelines are shaped by an advisory council of fourteen people from the insights industry and academia, and draws on a mapping of 34 synthetic data solutions from 30 providers, as well as academic research, and guidance from Esomar and the Insights Association.

Christina Habib (pictured), formerly of P&G, Unilever, Merck and Booz Allen, and a member of the advisory council, states: 'We are following those changes from the perspective of the practitioners who work with this data every day, and what we publish will be revised as the practice matures.'

The other council members are:

• Taylor Anderson, Senior Director, Global Insights Capabilities, Colgate-Palmolive
• Nima Gohil, Director, Market Consumer Intelligence, LVMH
• Gavin Johnston, Director of User Experience Research, Hyatt Hotels Corporation
• Christian Kurz, Senior Vice President, Global Streaming Research, Paramount
• Fenny Léautier, Head of Consumer Insights & Marketing Academy, Philips
• Dr Massimiliano Marinucci, Founder, NumiNate; formerly Coca-Cola and Unilever
• Rory McDonnell, VP Insights & Analytics, Global Marketing, Glanbia
• Michael Nevski, former Director, Global Insights, Visa
• Carl Persson, Vice President, Head of Global Food Insights, PepsiCo
• Prasad Polur, Partner, Suvan Ventures; formerly Unilever and Whirlpool
• Joni Salminen, Associate Professor, University of Vaasa
• Nils Stotz, Head of Product Experimentation, Zalando, and
• Olivier Toubia, Glaubinger Professor of Business, Columbia Business School.

The Six Types of Synthetic Data: A Field Guide for Buyers of Market Research is available now via the body's web site at www.syntheticdatastandard.org .