Hasdeep Sethi, group AI lead and data science director, Strat7
We’ll see more impact with conversational synthetic tools than quant surveys. Clients are already working with us on chatbots and synthetic personas, but these still rely on primary research – they are not a replacement. Quant is harder because surveys capture complex behaviour across multiple dimensions. Properly replicating that is going to take longer.

I think we’ll see improvements in 2026, but I’d be surprised if by December we are endorsing synthetic boosts across all our client work. The biggest potential is in consumer-facing, lower-risk categories (e.g. concept testing) where accuracy is already high. But highly regulated or technical sectors still need 100% human research.

Sabine Cronick, chief executive UK/EU, 2CV
2025 has shown some caution in this area, but I think we’ll see and hear more of it as time goes on. With low incidence audiences, sample quality concerns and a lot of budget pressures across the globe, it is inevitable that some clients will trial synthetic out to see whether it is ‘good enough’. Personally, I am still of the ‘human is best’ approach and would advocate for a smaller quality sample rather than larger ‘normalised’ data, but I do recognise for those under represented voices it can play a role.

Alexandra Kuzmina, innovation director, MMR Research
Synthetic data and audiences are definitely set to make waves in 2026. The appeal is obvious: speed, cost savings and the ability to simulate those elusive consumer segments.

But from a product experience research perspective, synthetic data isn’t a magic fix. Yes, it can help us stretch and enrich our datasets, but it doesn’t actually boost statistical confidence, it just gives the impression of it. The real difference will come down to how carefully these tools are validated and used. Synthetic audiences may be great for early ideation, but real progress still needs robust, representative research and expert judgement.

Sabine Stork, senior partner, Thinktank Research
I’m fascinated by synthetic data – it has huge potential not only in terms of budget-saving but also researcher creativity. At the same time there’s also huge potential to be misleading – AI has a tendency to be a bit glib, and synthetic responses may be overly neat, following logical paradigms which may seem appealing at first sight but which don’t represent the messy lives and thinking styles of real people.   

Suzy Hassan, managing director and co-founder, Potentia
Based on trials that Potentia ran comparing real data with synthetic data, we found that the data was so different that it wasn’t fit for purpose at the time. So based on our findings, there is likely to be an increased risk of higher levels of data scrutiny with synthetic data and audiences.

However, with the technology evolving rapidly, this will become more reliable and therefore provide a clear opportunity for using synthetic data to boost niche targets to help achieve more with less. A baseline of real data will always be valuable to support these synthetic models.

Jane Frost, chief executive, MRS
Synthetic data certainly has its uses and is providing exciting opportunities for the sector.  It’s important to remember, though, that synthetic respondents don’t exist in isolation; they are only made possible because of the high-quality real-world research that informs them. In other words, you can only use AI to predict what a certain group thinks about a topic because someone has asked them – in real life – something similar in the past.

The demand for regularly refreshed, high-quality data required to feed AI models will only grow as their use becomes more widespread – that’s why we launched the Campaign for Better Data last year, which we’ve seen the sector really get behind.

Kelly Beaver, chief executive officer UK and Ireland, Ipsos
Synthetic data has the potential to be quite a powerful tool within the industry, allowing us to scale research at more affordable costs to our clients, but it’s only as good as the data that gets put into it. We therefore need to be cautious about it when it’s applied so as to not replicate biases or distortions that often come from small or not sufficiently robust original samples.

Marie Ridgley, chief executive, UK insights division, Kantar
Done right, synthetic data could solve one of the sector’s trickiest challenges: reaching the hard-to-reach. It can scale smaller samples in longitudinal studies and open doors to underrepresented communities. 

To harness it properly, we need to double down on data standards. Synthetic audiences and AI models must be built on quality proprietary data. We’ve partnered with MRS on its Campaign for Better Data to keep standards front and centre – but the whole sector needs to rally. If we let fundamentals slip, we risk not just missing the opportunity but damaging the credibility and influence of our work.

Frédéric-Charles Petit, chief executive, Toluna
This is the single biggest shift in research, delivering unprecedented speed, quality and scale.

Babita Earle, international managing director, Zappi
Synthetic data will grow, especially for teams under pressure to reduce costs. But it’s not a silver bullet. Synthetic respondents can help validate assumptions, yet they’re only as strong as the most recent, high-quality data they’re trained on. Every idea you test with synthetic is being validated with yesterday’s consumer.

As an industry, we can’t lose sight of what matters: grounding decisions in high-quality consumer data. Over time, we’ll find the right role for synthetic inputs, but they should augment – never replace – real consumer understanding.

Will Ullstein, UK chief executive, YouGov
Synthetic data still faces deep scepticism in the industry. Its inability to capture real-world change makes it a poor substitute for high-quality research, and recent studies show outputs often lack the consistency needed for confident decision-making.

Instead of driving privacy-conscious innovation or broadening access, current applications risk misleading insights. In time, synthetic data may play a supporting role, but for reliable, high-impact research, data based on real responses from real people remains the gold standard.

Amanda Roberts, qualitative researcher, consumer strategy, Sky
I think more and more companies will start using synthetic data/audiences to emulate hard-to-access groups. But I do not necessarily think they should...

It is really important that we represent these people in our research, but I worry that we will take lower costs and higher convenience over biased results that lack depth, nuance and lived experiences.

Christopher Barnes, president, Escalent
Use of generic synthetic data will rise to a peak and fall quickly. Powerful, custom AI solutions that include synthetic will give client-side researchers more sway in organisations because they will provide speed and volume for the right populations in the right way.

Daniel Singham, commercial director, Yonder Data Solutions
The industry remains in an experimental phase with synthetic data, recognising its potential for delivering faster project timelines, reduced costs and more nuanced segmentation. However, ensuring high standards of data quality and accuracy is essential for successful adoption.

As 2026 progresses, I expect synthetic data and audiences to advance significantly, and the sector is likely to develop a clearer perspective and best practices around its use.

Matilda Andersson, managing director, Truth Consulting
I hope [synthetic data is] used more on the activation side – testing, exploring and pressure-testing ideas – rather than being treated as a replacement for real understanding of people, or decision-making.

Nick White, head of strategic research, Attest
We need to separate synthetic personas from synthetic panels. Synthetic personas will have the bigger near-term impact as a fast, early testing ground for creative, concept and strategic ideas. They’ll speed up decision-making and free up budgets for higher-value projects like segmentation, market sizing and conjoint.

Synthetic panels will still be in test-and-learn mode, moving from broad, averaged results toward greater nuance and realism. 2026 will be about refining these models before they become mainstream decision tools.

Ray Poynter, managing director, The Future Place and co-founder, ResearchWiseAI
It will grow steadily through the year; it won’t displace anything in 2026, but it will grow.

James Endersby, chief executive, Opinium
Before synthetic data can transform the industry, it needs to be handled with care. Rigorous ethics, transparency and validation against real-world benchmarks are non-negotiable. It should never replace fresh, unbiased primary research; it’s a tool for early-stage exploration, not final decisions.

Used responsibly, it can speed up innovation, fill gaps in underrepresented demographics and protect privacy. Done right, it’s a powerful complement to traditional research, enabling speed and inclusivity without compromising integrity.