• by Laurie Sullivan , 2 hours ago

Data accuracy as it relates to agentic commerce has become a hidden crisis for brands as retail heads into the holiday season.

There is a deep fear associated with the amount of synthetic content that Caroline Giegerich, vice president of AI at the Interactive Advertising Bureau (IAB), sees across the industry. 

“If I were a brand marketer today, I would think about it constantly,” Giegerich told the audience at Emarketer’s agentic AI conference on Tuesday.

Inaccuracy rates caused by AI shift between 10% and 20% across overall AI chat responses, said Alex Sherman, CEO of Bluefish.

“Those inaccuracy rates spike in heavily regulated industries like financial services, healthcare and pharmaceutical, which is the opposite of what should be true,” Sherman said.

Data scattered through open-and-closed web sources -- as well as advertising, paid and organic -- attracts agentic bots that scrap information from wherever the technology can find it. Sometimes that information is inaccurate.

“All these agents have different diets,” said Molly Schonthal, managing director of agentic commerce at WPP Enterprise Solutions. “They are gulping up different bits of information about your brand. Some rely heavily on what’s on a brand’s website, or within the four walls of Walmart.”

The lights are on as humans read a webpage, but when the lights go off and agentic agents interpret the brand, it likely will not interpret the same information on the page as a human.

Agentic agents look at data differently. They will read and interpret words like “bold” differently, and website content and the words used to describe products, as well as how that content is organized on the website.

The brand may list coffee as containing zero milligrams of caffeine in one place on the website and one milligram in another. The product may even be listed on a retail site citing a different amount of caffeine.

Social media and earned media are other aspects of a brand’s data footprint. “Think of the sum total of your brand’s digital presence as a data surface that’s consumed by an agent,” Schonthal said.

Schonthal asks when an agent “eats a brand’s data” to decide on behalf of the consumer what it needs to know in order to make a valid choice.