Two Major AI Launches for NIQ

Consumer intelligence giant NIQ has launched BASES AI, which will tap consumption pattern data from hundreds of thousands of households to mimic consumer behavior; and a CPG-specific LLM, which will fuel the advanced product segmentation capabilities of its flagship Discover retail measurement platform,

Evolved from the Buy division of the old Nielsen, and having recently combined with GfK, NIQ employs around 2,000 engineers and analyses around $7.2 trillion in annual global consumer spending, with data from more than 3,200 retailer partnerships. It also provides reports on consumer trends including GenZ habits bulletin SpendZ and regular Consumer Outlook reports analyzing trends, behavior and sentiment.

BASES AI, currently rolling out to beta-testers in Morth America, pulls in consumer-permissioned data in real-time to create large panels of synthetic respondents, which it says will 'replicate trusted market research techniques and methodologies' in a fraction of the time - ten minutes instead of 'days or weeks'. NIQ suggests that CPG companies can use these scores, together with 'rich guidance on drivers of interest and uniqueness', to develop ideas ready for final testing with human consumers. One benefit will be the potential to test larger numbers of ideas.

The LLM (large language model) makes use of NIQ's database of 160 million products, and according to the company outperforms public LLM models sevenfold in its ability to quickly and accurately code items. NIQ says the model will continue to evolve in both capability and sophistication.

Ramon Melgarejo (pictured), the company's President, Strategic Analytics & Insights, says the releases are 'the latest examples of how we are deploying our analytic horsepower across market-leading data sets to generate unparalleled insights that help CPG companies gain market share globally'.

The firm is online at www.niq.com .

All articles 2006-23 written and edited by Mel Crowther and/or Nick Thomas, 2024- by Nick Thomas, unless otherwise stated.