Originally published: AdNews, 19 November 2025
Young Ham, Global Head of Innovation & Partnerships, Pureprofile.
Surveys have long been the rightful foundation of both traditional and modern market research, offering a simple, effective, and straightforward way to understand consumers. However, as technology supercharges specialisation, niche respondent pools become harder to penetrate. As a result, data gaps are increasingly leading to delays, increased costs and incomplete insights that can impact time-sensitive research. The humble survey doesn’t need to be reinvented; it just needs a complementary ally.
Synthetic data has emerged as an alternative solution. Various options abound, a viable approach is engineered off real data samples and real-world personas. It complements traditional research and lets brands make faster, smarter decisions, acting as a “top-up” alongside traditional panels.
Bridging the gaps in a fragmented audience
Today’s audiences are highly fragmented across different channels and platforms. There’s frequently a lack of reliable data from underrepresented audiences, making it harder than ever for researchers to capture an accurate picture of consumer behaviour. Weighted data is functional, but lacks the extra raw responses as it models to amplify the results.
Another issue is how rapidly consumer preferences shift, particularly among younger generations influenced by ever-changing, ephemeral trends on social media. Even if a sufficient respondent pool is available, by the time a traditional survey is completed, sentiment may have morphed or evolved.
Marketers need faster and more adaptable ways to understand how their customers think and act. Synthetic data offers an immediate solution to fill data gaps and ensure complete, representative data collection without compromising quality or breaking the budget. Instead of waiting days or even weeks for research results, marketers can simulate reactions instantly, accelerating decision-making.
How synthetic data works
Synthetic data is not “fake.” It encompasses a range of approaches, such as digital twins that replicate real human respondents to represent survey populations, synthetic personas that model specific audience segments, and AI-powered statistical models that simulate human behaviours. Together, these methods allow researchers to generate representative data without relying solely on traditional surveys.
For marketers, it’s akin to assembling a rehearsal audience to predict how real customers may react. Synthetic data offers a way to amplify insights, generating personas and modelling audience responses far beyond a small sample size. Brands can test ideas much more quickly and unlock insights that small sample sizes can’t deliver. It’s important to remember the objective – what problem you’re solving.
Testing with less risk
Synthetic data is transforming marketing by letting brands experiment boldly, explore niche audiences, and test creative ideas rapidly, all without putting real people or reputations at risk. Here are three areas where it’s making the biggest impact:
- Creative prototyping: marketers can test multiple campaign variations in synthetic environments before investing in costly real-world rollouts. If you want to analyse twenty different ads but only have capacity to test a few with real audiences, deploying first with synthetic respondents to evaluate to narrow down to the few ads for real audiences to complete the assessment in determining the winning ad is an option.
- Niche markets: if the target audience is under-represented, synthetic modelling can amplify limited insights by boosting the sample size. As an example, you may be seeking opinions from 1,000 respondents from various demographics. The more niche you get, it’s more difficult to find the right candidates. If you need 100 in a particular group and can only find 80, you can model off the 80 collected responses to augment 20 extra cases.
- Sensitive topics: when testing campaigns on delicate issues and commercially sensitive projects, synthetic personas let marketers experiment safely without reputational risk. If a brand wants to keep a new concept under wraps, it may not want real people seeing it. By creating a synthetic audience, you can test ads and gauge how real people would respond to that message.
The challenges: trust and variability
While synthetic data offers great promise, it has its own challenges. Human behaviour is ultimately unpredictable and no machine can perfectly capture that randomness.
Synthetic data can also be inconsistent. The same model may generate different answers with each run, making repeatability tricky.
Freshness also matters. If the underlying data isn’t frequently updated, synthetic insights quickly lose relevance. Data hydration matters as synthetic models are only as accurate as the freshness and completeness of their underlying datasets.
A complement, not a replacement
Ultimately, synthetic data is only as strong as the data it’s trained on. Synthetic data can’t replace traditional data from live humans, nor is it intended to. Rather, it can be used to complement insights generation and must always be validated against real-world inputs.
At its core, synthetic data is predictive modelling based on the dataset it’s trained from, needing continuous fresh data to remain relevant. As such, it cannot fully replace existing data gathering methodologies, nor real human respondents. However, when combined with live surveys and behavioural research, it supercharges traditional methods, increases the efficiency and effectiveness of how data is used, and provides a more holistic view of audiences.
Young Ham
Young has close to two decades of experience in the market research industry, having worked for companies including Kantar, Lightspeed Research and TNS. Most recently he held the role of Director of Data Solutions & Sourcing at Kantar, where he led the data products and sampling partnerships for the APAC region.
Young is a qualified PRINCE2 Practitioner and has also completed his Lean Six Sigma, Green Belt training.
