Major brands are introducing AI into their experiences, offering customers new options for everything from product discovery to customer service.
Just this week, Dick’s Sporting Goods announced that it will add conversational AI experiences to the Dick’s app, while Ulta Beauty launched an AI shopping assistant, and The Home Depot officially began rolling out an AI voice assistant for in-store customer service calls.
These are just a few of the many recent CX-focused features companies have rolled out this year. However, a question remains: Are consumers interested in using these options over more traditional channels?
The answer depends on how the application fits a customer’s needs and habits, which also shapes how businesses should approach first-party AI, according to experts.
Teams will want to roll out AI concierges with a defined purpose, prepare for customers and others to try to break them, and understand how consumers view first- and third-party platforms if they want to make the investment worthwhile.
Even when AI is implemented well, trust remains a barrier. Consumer adoption of AI is uneven, and shrinks as they get deeper into a transaction, according to Chuck Gahun, leader in Forrester's digital business and strategy practice.
Forrester research has found that 85% of consumers say they find responses from AI answer engines helpful, but only 71% say they trust the provided answers, Gahun said. Few transactions take place in AI app experiences.
That doesn’t mean AI has no place in a digital brand experience. The technology has opportunities to improve the experience by meeting customers where they want it on their journey.
AI features need a purpose
A foundational consideration for leaders is whether they’re adding AI features because they serve a purpose, or because they are pursuing a buzzword. One approach will lead to better outcomes than the other.
“The way that brands have been successful in doing this is not to think about this as a one-size-fits-all and just turn on an AI app in Adobe, Salesforce or a customer-facing application,” Gahun told CX Dive. “Where businesses are being successful with this is being very deliberate about what value they're trying to gain by targeting very specific use cases.”
Use cases can vary industry by industry and company by company, according to Gahun. A hospitality brand may want to help customers discover the right vacation package, while a retailer’s AI could help them navigate the returns process.
AI’s trust problems stem from when the technology has gone wrong in very public ways, Shashi Bellamkonda, principal research director at Info-Tech Research Group, said. People are skeptical, and it's up to the brands to prove their AI implementations are useful.
“AI should not be gimmicky,” Bellamkonda told CX Dive. “It should have actual value to both the company and the customers. The best AI is used where you and I don't even know that AI was used.”
Consumers have been interacting with AI for longer than they realize, Bellamkonda said. Google has been experimenting with AI in its search engine for over a decade, and search continues to be a top use case for an application of AI that enhances the customer experience.
Expect people to try and break it
It’s no secret that guardrails can play a role in successful AI implementation, whether the technology is in a consumer-facing role or behind the scenes.
Businesses should limit AI to answer questions within the exact purview they intended, according to Bellamkonda. AI, unlike a human, doesn’t know when to ignore absurd requests. The more leeway the technology has, the greater the opportunity for problems to arise.
“Where things have gone wrong is if the implementation is not specific to their database,” Bellamkonda said. “People are going to try to break your AI tool assistant or app by asking questions.”
One example is Taco Bell’s AI drive-thru system, where a customer seemed to crash the system by ordering 18,000 cups of water. The problem, and the resulting viral social media clip, illustrates why businesses need to include clear guardrails when they implement AI, according to Bellamkonda.
Customers looking to cause a little bit of mischief aren’t necessarily the only stress test an AI agent will face.
Businesses need to be wary of potential security vulnerabilities third parties could exploit, according to Gahun. For example, someone could feed an AI concierge malicious input that can go on to corrupt some of the data within the organization.
“These are some of the topics that are coming up on agentic security in general that we're starting to focus on,” Gahun said. “I think businesses should be pretty cognizant and aware of what they're doing before they start deploying in-app agents.”
Consumer trust varies depending on the context
Consumers’ trust in an AI concierge depends on ownership, regulations and how the technology is being used.
Consumers tend to trust first-party owned experiences over third-party platforms like ChatGPT, according to Gahun. They are much more comfortable using third parties for product discovery, and then visiting a brand-owned app or website to complete the transaction.
The transition from an AI answer engine to a brand website can cause friction. Traffic from third-party platforms has been dropping across multiple brands, Gahun said. However, that doesn’t tell the whole story.
“That is not correlating to fewer conversions,” Gahun said. “Which means that the consumers that are coming through and basically go from ChatGPT to the brand-owned environment are converting, essentially, at a higher rate.”
This behavior is in line with how consumers are using AI in general, according to Gahun. They are much more likely to take the suggestions and log into their usual account with a brand to make the transaction.
For now, consumers are wary of entering their credit card information on platforms like ChatGPT, according to Bellamkonda. This isn’t necessarily a problem — companies are best off investing in ways that allow people to shop how they would like, rather than try to force a new paradigm.
