(Bloomberg) -- Consumer signals gathered using generative artificial intelligence for more than 240 million US adults detected heightened reluctance to spend last month in an economy that otherwise remained strong.

An index launched Monday by marketing automation software firm Zeta Global Holdings Corp. tracks people’s activities online and offline, gauging their interest and intention to buy in sectors including automotive, dining, financial services, health care and travel. 

By analyzing consumer behavior using computer language models, the Zeta Economic Index aims to assess the state of the economy and forecast trends in real time. The index is made up of two components: a score on the health of the economy and a measure of how well consumers can withstand downturns. 

“We’re actually looking at how much money they’re spending, we’re looking at what they’re researching before they spend it, and we’re able to ingest all of that information into our algorithms and layer it on top of the other normal data points and really come up with a better sort of forecasting,” David A. Steinberg, co-founder and chief executive officer of Zeta Global, said in an interview.

The gauge, partly based on data from 6.2 million publisher sites, detected a shift toward more cautious spending last month in an economy that is still showing steady progress, Steinberg said. 

The findings seem to align with mixed signals in the latest government data. While some parts of the economy have been losing steam this year, consumption adjusted for inflation rebounded in May after a pullback in April.

The Zeta Global index adds to the multitude of alternative data now available to hedge funds, private equity firms, asset managers and policymakers alike as they assess the health of the economy. Those datasets are often more timely than government reports and advances in AI technology provide an another way to leverage insights on consumer behaviors and economic activities. 

Zeta Global plans on updating the index monthly and has compiled several years of historical data.

(A previous version of the story was corrected to remove details on sectors in the fifth and seventh paragraphs.)

©2024 Bloomberg L.P.