Brand chatbot ROI falters as third-party AI usage surges 2x in one year

SaaS providers selling AI-powered customer service solutions are facing an existential challenge: their clients' own customers prefer not to use the branded chatbots they've built. According to Gartner, usage of these tools has flatlined since 2022, while third-party AI usage has skyrocketed, suggesting that the market may be moving away from siloed enterprise chatbots toward universal AI assistants.

A new Gartner study has revealed a stark disconnect in how consumers are using AI for customer service. While the use of third-party generative AI tools like ChatGPT, Claude, and Copilot has doubled in the past year, engagement with brand-deployed chatbots has remained statistically flat since 2022. This gap has significant implications for businesses that have poured billions into custom AI chatbots, only to see customers bypass them in favor of tools they already trust from their personal lives. According to Gartner analyst Keller, two-thirds of consumers now use generative AI in some form, but the vast majority of that usage is through third-party platforms, not brand-owned assistants. The data points to a fundamental mismatch between corporate AI investment and actual consumer behavior, leading to disappointing return on investment for many customer service initiatives.

“

While the use of third-party generative AI tools like ChatGPT, Claude, and Copilot has doubled in the past year, engagement with brand-deployed chatbots has remained statistically flat since 2022.

Keller emphasizes that simply adding AI to an existing, underutilized chatbot is unlikely to drive engagement. 'If customers are not already engaging with your chatbot, simply putting AI in that chatbot is probably not going to drive more engagement,' he said. 'So you really need to drive intentional adoption strategies.' The research highlights two critical areas where brands are falling short: taking action and presentation. Over 58% of consumers have used generative AI to complete a task, and nearly three-quarters of B2B customers have used it to accomplish a task, indicating that users want AI that not only answers questions but executes outcomes — such as processing a return, changing an order, or troubleshooting a technical issue. Brand chatbots, often limited to scripted responses and narrow knowledge bases, struggle to match the fluid, multi-step reasoning capabilities of foundation models.

The trust factor cannot be overstated. Consumers increasingly rely on AI tools like ChatGPT and Claude for everything from travel planning to medical queries, building a comfort level that brand-specific bots cannot replicate. 'They use these tools in every aspect of their life, and they increasingly trust those tools and trust the responses that they get,' Keller noted. When a customer service issue arises, the instinct is to open a familiar interface rather than search for a website's chatbot widget. This behavioral shift is poised to reshape customer service architecture. Brands may find themselves no longer in control of the initial support interaction, as third-party AI becomes the new front door.