Dose, a fast-growing consumer brand selling through DTC, Amazon and Retail, wanted to know ChatGPT Ads' true contribution to new-customer acquisition before deciding how much further to invest. With WorkMagic's geo incrementality testing, the team learned that click attribution was undercounting ChatGPT's real impact by 2.3x and found out about halo effects on Amazon orders. That clarity gave Dose a concrete, evidence-based plan for continuing to scale.
Key Takeaways
With WorkMagic, the advertiser’s team gained the following insights and growth:
2.3x more incremental DTC orders than click attribution reported
67% of incremental orders coming from new users
34.5% halo effect on Amazon orders
When you’re making acquisition decisions against a strict CAC target, having an incomplete view of a channel can lead you to make the wrong investment decisions. Incrementality helped us see the demand ChatGPT was actually creating, not just the conversions it could claim, and that’s a much better foundation for deciding how aggressively to scale.
Dose
Dose is a clinically backed, doctor-formulated, liquid supplement that delivers measurable results for the core systems quietly powering your body: from cardiovascular to metabolic health. Dose sells direct-to-consumer through Shopify and also lists on Amazon and at select retailers, giving the brand three sales channels to manage acquisition across. Growing new-customer volume efficiently on both fronts is central to the team's paid media strategy.
Proving ChatGPT Ads' real contribution to new-customer growth
Dose had started testing ChatGPT (OpenAI Ads) as an acquisition channel but couldn't yet tell how much of its performance was incremental. Click-based attribution can either overstate or understate a new channel's impact, and with new-customer economics under a strict CAC goal, the team needed a defensible answer before committing more budget.
The question was specific: what is ChatGPT's true incremental contribution to orders and new customers, across both of Dose's sales platforms — DTC and Amazon — not just the orders a last-touch model happened to credit it with.
A geo lift test measuring ChatGPT at the platform level, across both sales channels
WorkMagic and Dose designed a Pause-to-Measure geo holdout test focused on ChatGPT Ads at the platform level. Using machine learning to match Designated Market Areas (DMAs) on factors like seasonality and performance trends, the test held out a selection of U.S. DMAs as a control group while the remaining markets kept ChatGPT Ads running as normal.
Critically, the design measured impact on both of Dose's online sales platforms — DTC (Shopify) and Amazon — in order to detect any halo effect directly caused by ChatGPT Ads exposure.
Validated new customer contribution; bigger impact than attribution shows
Click attribution undercounted ChatGPT's real impact by 2.3x
The geo lift test measured 2.3x more conversions than click attribution previously reported. That gap is a direct illustration of why attribution alone couldn't answer the original question.
67% incremental orders from new customers
When looking at the distribution of incremental orders across existing and new customers, the test showed that 67% came from net new customers to the brand, demonstrating ChatGPT Ads’ value as a new customer acquisition channel with a distinct audience.
Amazon halo pulled blended CAC down
When combining ChatGPT Ads impact on Amazon in addition to DTC sales, the test proved ChatGPT Ads’ multi-channel incremental contribution to overall sales.
With a clear, incrementality-based read on ChatGPT's performance, Dose is well-positioned to continue refining targeting and creative strategics to ensure alignment with CAC goals, keep evaluating the channel on its full-funnel, multi-touch contribution rather than last-click credit alone, and eventually rerun WorkMagic’s media budget optimizer with ChatGPT included to find its efficient scale ceiling.