Nielsen Sports SA puts 150 million sports fan data points behind AI platform

Nielsen Sports SA has launched SIYA (Sponsorship Intelligence Yielding Answers), an AI-powered platform designed to give brands, broadcasters, rights holders and agencies faster access to sports fan insights without having to commission research or wait for an analyst to interpret the data.

The platform gives users access to more than 150 million sports fan data points through natural-language prompts, allowing them to ask questions about South African sports audiences and receive answers within minutes.

According to Tumelo Selikane, managing director of Nielsen Sports SA, the platform addresses a gap between the amount of sports research available and the time and expertise required to access it.

“The insight already existed, but it was locked behind time and specialists,” he says. “A brand, broadcaster or rights holder with a question about South African sports fans had to commission a study or wait for an analyst to pull it, which could take days.”

SIYA is designed to make that process self-service, allowing users to interrogate Nielsen Sports SA’s existing fan research when they need an answer.

Data, not AI assumptions

With growing concerns about the reliability of AI-generated information, Nielsen says SIYA does not generate statistics. Instead, it retrieves data from its verified research using governed queries.

“SIYA does not generate numbers; it retrieves them,” says Selikane. “Every figure comes from a governed query against our actual Nielsen Sports SA survey data.”

The platform applies the same rules an analyst would, including ensuring that only respondents who were genuinely asked a particular question are counted and identifying whether figures are weighted or unweighted.

“The AI interprets the question and presents the answer in plain language, but it cannot invent data,” Selikane says. “When a question falls outside what the underlying research supports, SIYA states that rather than guessing.”

The platform is also being developed for multilingual use. Users can ask questions in their own words, with Nielsen building support across South Africa’s official languages in phases.

Selikane says the challenge is not simply translating a question or answer, but ensuring that statistical meaning and caveats remain intact.

“The genuinely hard part is not translating the words; it is making sure a statistic and its caveats survive translation intact,” he says.

From research to decision-making

Nielsen expects the faster turnaround to change how organisations use research in sponsorship, audience and media decisions.

“When an answer takes days, you only ask the big, pre-planned questions,” says Selikane. “When it takes a minute, you can test an idea in the room, follow a hunch, and check three options before you commit budget.”

Early users have included sponsorship agencies, broadcasters, a global sports league, a major technology platform and brands outside traditional sports marketing, including education and FMCG.

The questions have also extended beyond the size of a particular fan base.

Users have asked about what fans drink, whether they bet, which platforms they watch on and which sponsors they already associate with teams, looking for potential commercial connections.

Some users have gone further, asking SIYA to help shape media strategies rather than simply provide audience insights.

Built in-house

Nielsen developed SIYA internally over the past year, with Selikane saying control over accuracy was a key reason for taking that approach.

“An off-the-shelf assistant is very good at sounding confident, including when it is wrong, which is the opposite of what a research business can put its name to,” he says.

Building the platform internally allows Nielsen to connect the model directly to its verified data, apply its own rules around bases and weighting, and retain control over client and respondent data.

SIYA is not intended to replace Nielsen’s analysts. Users can escalate complex questions to the company’s research specialists, with the platform handling “fast, factual and repeatable” questions while bespoke, strategic and methodological work remains with analysts.

“The aim is not to replace that expertise; it is to free it for the work only a person should do,” says Selikane.

Expanding beyond fan attitudes

Future developments will add total audience sports consumption and viewership data, giving clients greater visibility of how fans follow, watch and subscribe to sports across different platforms.

For broadcasters and advertisers, this could provide a more detailed picture of where sports audiences consume content and how that behaviour is changing.

Nielsen also plans to add verified sports sponsorship news, deal announcements and other proprietary research to the platform.

Further ahead, SIYA is expected to incorporate travel and tourism data, allowing brands and rights holders to explore how sports fans travel, where they stay and what influences their travel decisions.

“Ultimately, every piece of in-house research we conduct will feed into SIYA, creating a continually growing source of verified data and insight that connects fan attitudes, behaviours, consumption and commercial activity in one place,” Selikane says.

The platform is also designed to provide Nielsen with a clearer view of what the market wants to know. Questions submitted by users can highlight recurring topics and gaps in existing research, informing future Fan Insights waves and data collection.

As South African sports audiences increasingly move between linear television, streaming, social media and live events, Nielsen sees SIYA as a way of bringing those behaviours into a single view.

“Today a single fan might follow a team on linear TV, a stream, a highlights clip on social media, and sometimes the stadium, often in the same week,” says Selikane.

“SIYA is designed to describe that fan across all of those platforms at once, so an organisation can see the whole picture of how sport is followed now, instead of stitching together separate single-source numbers.”