WPP Media has partnered with Givsly to test synthetic audience segments that incorporate charity donation patterns for programmatic buys, Digiday reports. According to Tara Sadlak, group director of media delivery at WPP Media, early tests across four beauty and fashion campaigns produced a 2% lift in video completion rates; Sadlak declined to name the brands and told Digiday, "We're looking for discoverability and increased reach." Givsly CEO Chad Hickey told Digiday the system, built using the Claude LLM, maps U.S. zip codes with third-party sources including national census and election data plus nonprofit donation metrics to generate demographic, behavioral and values-based cohorts. Digiday reports WPP layered its own first- and second-party data into the synthetic segments for targeting.
What happened
WPP Media struck a partnership with Givsly, a tech company founded in 2019, to test synthetic audiences that include charity donation data as part of programmatic media buys, Digiday reports. Per Digiday, Tara Sadlak, group director of media delivery at WPP Media, said early tests across four beauty and fashion campaigns produced a 2% lift in video completion rates; Sadlak declined to name the brands and told Digiday, "We're looking for discoverability and increased reach." Digiday reports that Chad Hickey, CEO and founder of Givsly, described the system as generating audience segments containing demographic, behavioral and ethical reference points in response to advertiser briefs.
Technical details
Per Digiday, Givsly's system uses an LLM (Claude, from Anthropic) and creates synthetic cohorts by overlaying a U.S. zip-code map with multiple third-party inputs. Reported inputs include:
- •national census data
- •election results
- •nonprofit donation counts, frequency, and charity type
Digiday reports Givsly uses that composite dataset to infer proxies for disposable income and social values, then combines those synthetic segments with WPP Media's first- and second-party data for campaign targeting.
Context and significance
Public coverage has noted agencies have generally used synthetic audiences for planning or insight rather than as direct programmatic bid signals, because of hallucination risk and thin-data concerns, Digiday reports. The reported 2% video completion improvement in the WPP-Givsly tests is a concrete performance metric that the coverage highlights when describing why agencies may revisit that approach.
For practitioners, using donation-pattern proxies and geographic overlays is a variant of values-based segmentation that can augment demographic and behavioral signals. This approach trades on correlated public records to infer preferences where individual-level deterministic identifiers are unavailable or restricted.
What to watch
Observers should track three indicators: whether reported lifts replicate across categories beyond beauty and fashion; how vendors disclose data provenance and sampling density at zip-code level; and whether publishers and demand-side platforms accept synthetic cohorts as biddable segments at scale. Also monitor how vendors document chain-of-trust for third-party sources and any audit processes that guard against spurious or hallucinated attribute assignments.
Scoring Rationale #
A single-agency test of synthetic audiences using donation-pattern proxies with a modest 2% lift is a solid but niche adtech result - relevant to practitioners evaluating cookieless targeting alternatives, but limited in scope (four campaigns, beauty/fashion only) and single-sourced from Digiday. Does not represent a broadly validated industry shift.
Practice with real Ad Tech data
90 SQL & Python problems · 15 industry datasets
[Active Search Campaigns by BudgetEasy](/problems/sql/active-search-campaigns-by-budget)
[High CPC Clicks & Poor Landing PagesMedium](/problems/sql/high-cpc-clicks-poor-landing-page)
[Campaign ROAS by Attribution ModelHard](/problems/sql/campaign-roas-by-attribution-model)
250 free problems · No credit card