Ipsos’s Dr. Nikolai Reynolds: The future of R&D lies in AI powered by truth, ethics, and real consumer needs
Truthful, fresh, and relevant data will become the key differentiator as AI-saturated databases and models risk flattening and homogenizing insights, according to Dr. Nikolai Reynolds, Ipsos Global Head of Product Testing.
ADVERTISEMENT
Generative AI is rewriting the rules of product innovation. For Dr. Nikolai Reynolds, Ipsos Global Head of Product Testing, it is speeding up development cycles, cutting research costs by as much as half, and even creating market-ready product concepts in a single week. By blending synthetic and real consumer data, Ipsos achieves the same business decisions in 95 percent of cases, proving that AI can be both fast and efficient.
Yet, Reynolds is clear that innovation cannot rely on algorithms alone. Responsible use of AI means ensuring authenticity, preventing bias, and keeping human judgment at the center. As he puts it, the future of R&D is not just about harnessing AI’s creative and predictive power, but about anchoring it in truth, ethics, and real consumer needs.
In a conversation with Storyboard18, Reynolds discusses how client expectations around innovation are evolving globally, ways in which AI, Gen AI, and synthetic data are reshaping product development and consumer testing, and new capabilities that research teams need to build today to stay future-ready.
Edited excerpts:
How are client expectations around innovation evolving globally, and what does that mean for the future of research and development?
Globally, client expectations around innovation are shifting from traditional, slow processes to a demand for faster, smarter, and more agile development cycles. This evolution is propelled by the capabilities of Generative AI, which enables a move away from the conventional, siloed "stage-gate" innovation funnel. Clients now expect a more dynamic and continuous approach where insights are generated rapidly and are more closely aligned with authentic consumer needs. This shift fundamentally reshapes the future of research and development (R&D) into a model the document calls "Innovation Genesis."
This new framework replaces the linear, sequential R&D process with a connected, continuous, and seamless cycle. In this future model, elements of the product mix—such as ideas, concepts, and packaging—are developed simultaneously rather than in isolation. This is powered by a "consumer data fabric," which integrates diverse data sources to fuel both the generation and validation of innovations.
For R&D professionals, this means their role will become less tactical and more strategic. They will focus on evaluating the feasibility, viability, and scalability of AI-generated insights, ensuring the process is centered on the consumer. A critical component of this future is a commitment to responsible innovation, which includes ensuring data security, preventing AI biases, and maintaining human oversight to guide the technology ethically. This responsible approach is framed by the principles of Truth, Beauty, and Justice, ensuring that innovations are grounded in authentic data and protected from intellectual property risks.
In what ways are AI, Gen AI, and synthetic data reshaping product development and consumer testing—and how much can we trust these tools today?
Generative AI means AI that generates and covers the generation of synthetic data, texts, images, sounds and music. We use deep learning AI to generate synthetic consumers as they allow us to provide numerical respondent level data. Our validations have shown that with a part synthetic and real human sample, we achieve in 95% of the cases the same business decision as using a full human sample in product development at 20-60% reduced cost for the research as well as up to 50% more time. We use LLMs and Diffusion Models to generate product improvement ideas as well as generate positionings and improved or new packaging.
Our validations have shown that using GenAi we achieve product trial rates that are 10% higher than product trial rates where no GenAI has been used. Independent which GenAI model is being used (Deep Learning, LLMs or Diffusion Models) validations are important and using the data in safe and responsible ways are crucial. At Ipsos we have a safe and secure platform called Ipsos Facto which makes sure that no data is shared with the public models and each client specific data is not shared among clients.
What are some examples where AI or synthetic data has accelerated innovation outcomes or opened up new capabilities?
Across all studies, using synthetic data Ipsos allows to get clients to the same business decision in 95% of the cases as will a full human sample, but at 20%-60% of the costs and in up to 50% of the same time. Using LLMs and Diffusion models we help clients to achieve 10% higher Trial rates and 9% higher Overall Liking of the product vs not using GenAI.
To provide a concrete example: a standout example of AI accelerating innovation is the work done for a South African wine producer who faced tight deadlines for product improvement. The process began with an "Innovation Lab," gathering feedback from a small sample of 69 consumers. To accelerate the process and gain richer insights without recruiting more people, this small sample was augmented with 131synthetic users. This AI-driven technique created a robust, validated dataset that provided the confidence of a much larger study in a fraction of the time.
The most significant capability emerged from using InnoExplorer AI, a generative AI tool. Instead of just analyzing the consumer feedback, the AI was trained on the verbatim comments to creatively generate new solutions. It identified and quantified 10 specific areas for product improvement, ranking them by their potential impact on consumer liking. Based on the top-ranked improvement area—balancing sweetness and bitterness—the AI generated a complete, market-ready product concept named "Pinotage Harmony." This included a consumer insight, a clear benefit, and a reason to believe.
It even developed a new bottle design and packaging concept to align with the new product idea. This case is a powerful example of accelerated innovation because the entire cycle, from initial consumer testing to generating a validated new product concept with packaging, was completed in just one week. It demonstrates how AI can not only speed up research but also open new creative capabilities by transforming raw consumer feedback into tangible, market-ready outputs.
What new capabilities should innovation research teams build today to stay future-ready in the face of tech, consumer, and market shifts?
Databases will lose their effect more and more, as the data they capture and process will be AI based. In addition, LLMs quality will be suffering as the data that they trained on will be more AI based creating a flattening of insights and vanilizing insights as well as everybody will be using the same public information. Therefore, truthful fresh data will become more important and key differentiater as well as making sure the data is relevant for the target group.
