Goodway Group has signed a partnership with data company Arima to fold a “synthetic population” of U.S., British and German consumers into its media planning process, Digiday reports.
Synthetic audiences move from experiment to planning tool
According to Goodway Group CEO Paul Frampton-Calero, the agreement lets staff run incrementality tests in a simulated environment, then verify the results with small-scale media activations. Teams could model how awareness or share of voice responds at different spend levels without leaving a real market inert as a control.
Frampton-Calero said the approach helps fill “gaps” in limited traditional research, especially for particular products or regions. Goodway has worked with Arima since August and also uses Optimal to create synthetic audience personas. His stated ambition is to apply the tool beyond media buying to distribution and pricing questions.
Brands and agencies are already in
Goodway is not alone. Magnum, Vans and pharmaceutical advertiser Kenvue have worked with Vurvey Labs, which builds synthetic personas from public data and qualitative video surveys with real consumers. For Kenvue’s Aveeno Body, global innovation and brand communication lead Joy Doyin Ogunneye said synthetic research cut the time to write a strategy brief from 30 days to three days, and helped inform launch of Aveeno’s Sensitive Skin Scents earlier this year.
Beauty brand Sacheu has used Limitless Labs to create AI personas for virtual focus-group testing based on thousands of TikTok and Instagram videos. Earlier this year WPP began using a synthetic data solution built on charity donations data to inform U.S. media buys, while Dentsu has been working with startup Evidenza for more than a year. Code & Theory, which developed its approach with Adobe, has used synthetic personas for years.
Code & Theory’s Stef Hoffman said teams commonly use personas to guide creative development from pitch through production. The agency roots personas in client first-party data first, then layers third-party or public sources, keeping the model connected to real customers.
What to do about it
If you’re evaluating synthetic audience research, the practitioners quoted by Digiday point to two disciplines:
- Start with the right foundation. Julie Clark of TransUnion warned against models built on “garbage” data; Code & Theory’s first-party-first approach is designed to avoid that.
- Keep judgment in the loop. Hoffman put it plainly: “It informs judgment, but it doesn’t replace judgment.” That means validating simulated results with real campaigns before scaling.
- Test before you commit. Model spend scenarios, then confirm with small media activations.
“It informs judgment, but it doesn’t replace judgment.”
— Stef Hoffman, Code & Theory
Why it matters
Synthetic audiences promise speed and cheaper experimentation. Instead of waiting 30 days for a traditional brief, teams can query a simulated population and iterate quickly. But the risk is over-trusting models built on weak data. The agencies likely to benefit are those using synthetic research as a faster hypothesis engine, not a replacement for real market validation.
The spending backdrop
The synthetic push comes as ad budgets shift toward digital channels where modeling is easiest to apply. WARC forecasts social media ad spend will rise 21.3% to $394.6 billion this year, and global ad spend is projected to grow 11.9% to $1.34 trillion before slowing to 8.4% next year. In that environment, faster simulated testing can help marketers stretch tighter budgets.
Source: Digiday
