Third-Party Gen AI Doubles in Customer Service as Brand Chatbots Stall

What happened

Consumers are increasingly turning to third-party generative AI tools — think ChatGPT and similar assistants — to resolve customer service queries, and they are doing so at roughly twice the rate they were a year ago. Meanwhile, the brand-owned chatbots that retailers and service businesses have invested heavily in have seen virtually no statistically meaningful growth in adoption since 2022, according to research reported by Marketing Dive and Retail Dive.

The divergence is stark: third-party gen AI usage in customer service contexts has doubled over the past twelve months, while proprietary chatbots remain stuck at the same plateau they hit three years ago. Customers, it appears, are solving their own problems with general-purpose AI rather than waiting for brands to catch up.

Why it matters

For customer experience and service-design practitioners, this gap is a warning signal, not a footnote. When customers abandon a brand's own channel in favour of a third-party tool to answer questions about that brand's products or policies, the company loses control of the interaction — the tone, the accuracy, the upsell opportunity and, critically, the data. Every query resolved by an external AI is a missed moment of relationship-building and a potential source of misinformation about the brand's actual offer.

From a behavioural-economics standpoint, the pattern reflects a straightforward effort-reward calculation. Consumers have learned that general-purpose AI delivers faster, more conversational and more satisfying answers than the scripted, intent-tree chatbots most brands still deploy. Once a behaviour is rewarded consistently, it becomes habitual — meaning the longer brands delay improving their own tools, the harder it becomes to win customers back to proprietary channels.

By the numbers

• 2× — third-party generative AI tool usage for customer service has doubled in the past year.
• Since 2022 — company-provided chatbot adoption has shown no statistically significant increase over that period.

The Renascence take

Most brands will read this data as a prompt to upgrade their chatbot's underlying model. That is the wrong lesson. The real issue is not the technology stack — it is the experience philosophy sitting beneath it.

Brands have spent years building chatbots optimised for deflection — keeping customers away from human agents — rather than for resolution and delight. Customers have noticed, and they have voted with their fingers. The behavioural principle here is channel trust: once a customer decides a channel is unreliable, no amount of rebranding or model-swapping will recover that trust quickly. What customer-obsessed operators should actually do is audit every chatbot interaction for effort and resolution rate, kill the flows that frustrate, and be honest about where a human or a better-integrated AI genuinely serves the customer faster. Competing with ChatGPT on raw capability is a losing game; competing on contextual, brand-specific knowledge and seamless escalation is where proprietary tools can still win.