What Synthetic Audiences Are—and Why They Matter Now
As third‑party cookies fade and regulations tighten, advertisers are scrambling for privacy friendly targeting that still feels precise. Synthetic audience data is emerging as a powerful answer. Instead of tracking individuals across the web, AI models ingest large pools of first party data, panel responses and industry research, then generate statistically realistic “twin” groups that behave like real readers—but contain no personally identifiable information. News UK’s new Times ExplorAItion initiative shows how this works in practice. The Times and The Sunday Times feed stripped‑back subscriber behaviour, reader panels, engagement metrics and PAMCo data into Electric Twin’s synthetic‑audience platform. The result is virtual panels of high‑value readers—like high‑net‑worth individuals and business decision makers—that advertisers can query in seconds to test propositions, messaging and formats. In a market where CMOs must justify every pound of media spend, synthetic audiences promise faster insight with less risk to user privacy.
Inside Times ExplorAItion: From Subscribers to Simulated Segments
Behind the scenes, The Times has already been using synthetic audiences to steer its own business decisions, leveraging a base of 659,000 digital subscribers. Before launching a unified parenting channel, its teams used Electric Twin to test which topics felt missing, how younger, often female readers wanted the section framed and even how it should look visually. What once required weeks of qualitative research was compressed into rapid, AI‑driven exploration, grounded in real first party data rather than generic demographic profiles. The same synthetic audience infrastructure helped validate the importance of tackling password sharing and informed the roll‑out and communication of bonus accounts for premium subscribers. For advertisers, these AI advertising analytics tools function like a constantly refreshed, data‑fuelled focus group. Media planners can explore how Times readers move from print to apps, podcasts or YouTube, using media audience modelling to see where attention accumulates instead of relying on static reach and frequency charts.
From Demographics to Behavioural Twins: How AI Rewrites Targeting
Traditional audience planning leaned heavily on age, gender and income brackets. Synthetic audiences flip that model by starting with behaviour. AI systems digest vast repositories of first party data—what people read, how long they dwell, when they subscribe or churn—alongside panel and industry datasets. They then learn patterns that allow them to infer responses and create lookalike or synthetic segments that reflect real‑world behaviours rather than just static traits. For agencies such as Wavemaker, the appeal lies in turning sprawling datasets into a “living, breathing” picture of how audiences actually move across touchpoints. Synthetic audiences act as dynamic focus groups: planners can pressure‑test new creative, formats or content ideas daily, instead of waiting for quarterly surveys. That gives brands faster feedback loops and publishers new ways to monetise their data, potentially commanding higher CPMs on quality, evidence‑backed segments without exposing individual identities or relying on third‑party cookies.
Meta’s AI Pivot: Automation, Analytics and Platform Power
The same AI forces reshaping publishers are also transforming tech giants. Meta is reportedly planning to lay off around 8,000 employees and freeze hiring for roughly 6,000 roles as it doubles down on AI and reshapes its workforce. Internally, AI systems are already handling tasks such as coding, content creation and data analysis, reflecting a broader industry trend where machines increasingly perform work once done by humans. This shift is not unique to Meta; companies like Microsoft and Amazon are moving in a similar direction, and tens of thousands of tech jobs have been affected. At the same time, Meta is investing in roles for AI engineers, data scientists and machine learning experts. For advertisers, these changes signal a future in which platform‑level AI does more of the heavy lifting in audience discovery, measurement and optimisation—creating synthetic or inferred segments automatically, and giving platforms even more control over how audience data is packaged and sold.
Benefits, Risks and What It Means for Malaysian Brands
Synthetic audiences promise a compelling trade‑off for marketers and publishers in Malaysia and beyond. They unlock privacy friendly targeting built on consented first party data, offer richer behavioural insight than blunt demographic buys and can improve measurement by isolating which environments and formats genuinely drive attention. For premium publishers, this form of media audience modelling can support stronger pricing on high‑quality, verifiable segments at a time when cookie‑based retargeting is disappearing. The risks are equally real. If underlying datasets are skewed, models can encode bias and over‑represent certain groups. Lack of transparency about how segments are built can make it hard for brands to trust or troubleshoot performance. Over‑fitting is another danger: synthetic audiences that perfectly mirror historical behaviour may fail in new market conditions. Malaysian advertisers and publishers should treat these tools as decision support, not oracles—insisting on clear methodologies, regular back‑testing against real campaign results and human oversight over how AI‑derived segments are used.