Ideally’s Pitch: Move Research Upstream to Fuel Creative Risk-Taking
Market research firm Ideally wants to upend how the creative industry uses consumer data. Instead of commissioning studies that arrive after a campaign is half-built, the New Zealand-founded platform is pushing brands and agencies to bring synthetic audience feedback right to the start of the creative process. The idea, says co-founder and chief revenue officer Joshua Nu’u-Steele, is to give creatives a risk-reducing sounding board at the moment they hit a fork in the road, not a pass-fail judgment after the fact.
Ideally launched in 2023 as generative AI began to take off, and it has since built a platform where different teams can run consumer tests that compound over time. Data from one test can inform the next, creating a web of insights that a layer of AI uses to refine its output. The company has now landed its first end-to-end network partnership: a multi-year deal with Omnicom Oceania that gives every agency in the group access to overnight, targeted research results at roughly one-tenth the cost of a traditional research budget. Omnicom Intelligence lead Daniel Shaw says the value lies in moving from “I think” to “I know.”
Ideally’s ambitions are growing beyond Australia and New Zealand. The firm has opened a New York office, and its U.S. business has jumped 400%, with early traction in wine (Treasury Wine Estates is a foundation customer) and functional food and beverage. This week it also launched Research Assistant, an AI feature that turns a plain-language question or creative brief into a full study framework in a fraction of the usual time.
Where Synthetic Insights Fit — and Where They Don’t — in the Agency Landscape
The promise of synthetic data in advertising is seductive: faster, cheaper and — if the models are good — continuous. Ideally’s platform aims to make consumer research a living part of the creative workflow rather than a one-off checkpoint. The compounding-data feature means that the more an agency uses it, the smarter the system becomes about its brands and audiences. For Omnicom, embedding Ideally across 23 brand categories is an experiment in turning iterative testing into a shared intelligence asset, which could lower the cost of creative risk and speed up campaign development.
The Omnicom Integration: Scale and Caution
Omnicom Oceania’s deal shows how a holding company can operationalize synthetic research at the network level, plugging it into tools like Acxiom, RealID and Flywheel. The claimed one-tenth cost and overnight turnaround are real attractions when marketing budgets are tight. But the reliance on AI-generated audiences does not eliminate the fundamental challenge of synthetic data: most models trend toward the mean, smoothing out the outlier insights that sometimes power breakthrough creative. The platform’s internal research frameworks and human oversight are supposed to guard against model drift, but the industry’s skepticism remains live.
What the Skeptics Say
Nathan Richter, senior partner at Wakefield Research, captures the core concern. Synthetic research can inform early-stage thinking, he argues, but it should not drive final creative decisions because AI-generated audiences only replicate past human behavior; they cannot fully uncover genuinely novel human responses. When used in isolation, synthetic data risks recycling known patterns rather than sparking the kind of counterintuitive insight that produces a memorable campaign. Ideally’s counter is that its tool is designed to be used as a “gut-check” amplifier, not a sole decision-maker — a difference that will matter greatly to agencies weighing the cost-speed-quality triangle.
What the Omnicom Deal Means for Agencies and Marketers
- For holding companies and agency networks: The Omnicom Oceania model — platform access across all agencies with compounding internal data — may become a template for cost-efficient insight capture. Evaluate whether a network-wide synthetic tool can sit alongside your existing primary research rather than replacing it, and identify creative workflows where early-stage testing would produce the most value per euro or dollar.
- For creatives and strategists: Synthetic feedback at the “fork in the road” can help validate hunches and trim obviously losing directions earlier, but treat the output as directional, not deterministic. Pair it with even a small qual sample on make-or-break decisions to avoid the regression-to-the-mean trap.
- For the research industry: Expect synthetic-first platforms to keep pressuring pricing and turnaround times. Firms that offer hybrid models — AI speed plus human analytical depth — will likely defend the premium end. Watch Ideally’s expansion in wine and functional F&B for signs that category-specific synthetic models can build proprietary advantages.
Risk & Opportunity Assessment
| Commercial Risk | Medium | Ideally’s model depends on agencies and brands trusting synthetic data enough to embed it into early creative stages. If early adopters like Omnicom Oceania find that campaigns underperform because of overreliance on regressive AI insights, renewals could stall. |
| Competitive Risk | High | A cluster of well-funded AI research platforms and traditional firms launching their own synthetic tools could erode Ideally’s first-mover advantage, especially if incumbents integrate synthetic features into existing agency software ecosystems. |
| Regulatory Risk | Low | Synthetic data that does not use real personal information faces relatively low immediate regulatory risk, but any move toward personalization or merging synthetic outputs with actual consumer profiles could trigger data-protection scrutiny. |
| Reputation Risk | Medium | If a high-profile campaign fails in the market after being shaped by synthetic insights, the ‘AI led us astray’ narrative could damage both Ideally and the agencies that relied on it. The company’s messaging that it is a gut-check tool rather than a decision-maker helps, but the optics matter. |
| Technology Disruption | High | Rapid advances in generative AI may soon allow agencies to build their own bespoke synthetic research environments in-house using open models, bypassing dedicated platforms like Ideally. Ideally’s defensibility hinges on the compounding-data network effect and the embedded research frameworks. |
| Commercial Opportunity | High | The Omnicom partnership provides a scalable proof-of-concept that can be replicated across other holding companies. The 400% U.S. growth and early category focus (wine, functional F&B) suggest a path to industry-specific dominance if the product delivers measurable creative effectiveness improvements. |
Comments 0