Experian EVP: Lessons from Launching Apps in ChatGPT

TLDR

• Experian is treating ChatGPT as an emerging distribution channel, building capabilities that can also be deployed through its own AI assistant and potentially other platforms.
• The company is testing what belongs inside ChatGPT versus Experian, balancing fewer steps for consumers against privacy, data protection and regulatory requirements.
• Early testing has already forced design changes, and Experian is tracking engagement and conversions to determine whether ChatGPT ultimately warrants continued investment.

Experian’s move into ChatGPT is less a bet on one chatbot than preparation for a broader shift in how consumers may find financial products.

The data and credit reporting giant launched an insurance app in ChatGPT in February, added personal loan shopping in June and followed with a credit card app in August. Together, the releases show Experian testing whether external AI platforms can become viable distribution channels.

“We want to meet consumers where they’re already doing the research,” Debbie Hsu, executive vice president of product at Experian Consumer Services, said in an interview with The AI Innovator. “We see from our data that consumers are starting to use these assistants more and more.”

Consumers like the ease of asking a chatbot to find credit card offers for them rather than doing their own searches and filtering for preferred perks such as airline miles or cash back on purchases.

That mindset shift creates both an opportunity and uncertainty for companies like Experian accustomed to reaching consumers through search engines, email, apps and their own websites.

ChatGPT could help Experian reach a younger generation of AI-native consumers. But the company must operate within OpenAI’s evolving rules, including a requirement that users first install the apps in ChatGPT before they can be used.

Hsu compared the moment to the early days of search engine optimization. Companies did not know exactly how search rankings would evolve, but they recognized the value of establishing a presence early.

“We just want to make sure we’re there,” Hsu said.

Building for more than ChatGPT

Experian’s preparations for this fast-changing channel begin with its technology architecture.

Rather than build services exclusively for ChatGPT, the company is developing capabilities that can work across multiple AI interfaces, including Eva, Experian’s virtual assistant. Personalized credit card offers, for example, are already available through Eva.

The goal, Hsu said, is to be “building once and deploying in multiple places.”

That approach gives Experian some protection if consumers gravitate toward a different AI platform – or if ChatGPT never becomes a significant source of traffic. Experian is evaluating each AI service on its own merits.

The company has not launched a comparable consumer experience on Anthropic’s Claude because it currently views Claude as more focused on enterprise users, Hsu said. Google’s Gemini supports connected apps, but she described the broader market for such integrations as still developing.

Experian is therefore investing in AI distribution without making any outside platform indispensable.

“We’re building this not just for ChatGPT,” Hsu said. “Whatever we build, we’re also building it for our own AI assistant. That’s where the multi-surface strategy for us is beneficial. In that respect, I’ll say it’s never a waste.”

Because services developed for an outside AI platform can also be used through Eva or elsewhere, Experian does not currently consider its ChatGPT work a wasted investment even if adoption remains limited.

But that commitment is not open-ended. If ChatGPT does not eventually become more effective at helping users find relevant third-party apps and services, Hsu said, Experian could reconsider whether continued investment is worthwhile.

Building multiple versions

Experian is deliberately treating the apps as early product releases rather than trying to recreate its entire marketplace inside ChatGPT.

The credit card app displays offers from participating lenders after users tell ChatGPT their preferences. Hsu said Experian doesn’t ask for their credit score but rather it is self-reported. Users can compare details such as annual fees, introductory bonuses, rewards rates and other information.

Consumers who want credit card offers that match their credit profile are directed to Experian’s website, where the company can use its data about the user and lenders’ eligibility criteria to find matching offers.

While that extra step adds friction, Hsu said Experian balances the convenience of completing more of the transaction inside ChatGPT with the need for consumer privacy and data protection.

“It’s a business decision as well as a privacy and data protection consideration for us,” Hsu said. “It’s a balance across all of those dimensions.”

The balance could shift as Experian learns more about how consumers use the apps. Hsu described the current apps as “version one.” The longer-term goal is to make it more convenient for consumers to shop for offers while deciding what can be safely shown inside an external AI platform.

Experian signaled that direction when it launched its auto insurance app in February. Consumers start by entering their zip code into ChatGPT to get some initial offers, with a link directing them to Experian’s website for more personalized choices. The company said it planned to add deeper personalization over time in the app, including letting consumers choose and finalize their auto insurance offer all in ChatGPT.

The auto insurance app was temporarily unavailable at the time of Hsu’s interview because Experian was resubmitting it as part of what she described as a “major update.”

Rethinking the design

Building for ChatGPT has also forced Experian to rethink interactions originally designed for websites.

Experian’s website lets consumers narrow down their choices one step at a time by using filters. It initially wanted to reproduce that process in ChatGPT, but testing showed this didn’t work well, Hsu said. So Experian redesigned the app to ask for three pieces of missing information at once.

The change may seem small, but it illustrates why the company is taking an iterative approach. Moving an existing marketplace into an AI assistant involves more than transferring the same experience to a different screen.

Experian has been less willing to experiment with the underlying product information. Card descriptions, rewards details, advertising disclosures and other offer information must meet regulatory standards and lenders’ marketing requirements, Hsu said. Experian therefore keeps that material consistent across platforms.

The presentation, however, can change including whether an interface displays for example three offers or five.

The apps show products from Experian’s participating lenders, not every card or loan available in the market. Those lenders provide eligibility criteria Experian uses when generating personalized offers through its own services.

Hsu declined to discuss how Experian’s marketplace ranks offers. The company receives commissions when consumers accept credit card and loan offers. Insurance operates under a somewhat different model, she said, but Experian also earns revenue when consumers transact.

Measuring the AI channel

For now, Experian is evaluating ChatGPT as it would any emerging customer acquisition or referral channel.

The company is tracking app connections, the number of consumers who start conversations, how many turns those conversations involve, clicks to personalized offers and subsequent account creation or login. The ultimate measure is whether a consumer arriving through ChatGPT accepts an offer.

ChatGPT remains a small source of traffic compared with Experian’s established channels. Hsu said it “doesn’t even compare” with email yet.

For now, the company is learning how conversational commerce works while the rules governing app discovery, personalization and transactions on major AI platforms are still taking shape.

Experian has experimented early with emerging interfaces before. Years ago, the company built an experience for an early version of Amazon’s Alexa, Hsu said. It eventually shut down the service after deciding that voicing credit scores and other sensitive information aloud inside a home raised privacy concerns and made the experience a poor fit.

Whether ChatGPT develops differently remains an open question. Experian’s strategy is to keep learning and adapting.

What’s key is Experian wants its financial marketplaces to be available wherever consumers begin their research – while building the underlying technology so it can follow users wherever they eventually land.

“The more we show up, the better,” Hsu said.