Over the past two years almost every company has put some kind of AI into its customer service: a chatbot, automated email replies, a voice assistant. The money has been spent and the projects are live. The question owners and managing directors are asking now is different: did it pay off? According to a Gartner survey published on 4 November, in the vast majority of cases the answer is not yet.
What the Gartner survey shows
Gartner surveyed 275 customer service and support leaders between July and August 2026. The headline: just 4 percent of organisations have achieved the return on investment they expected from their AI initiatives. For the other 96 percent the technology works, but the business impact falls short of the plan.
The research singles out three things that separate what Gartner calls “AI-forward” organisations from the rest. First, they use AI to create value rather than only to cut costs: they analyse customer conversations and feedback at scale to find recurring friction and remove its cause. Second, instead of isolated pilots they build a portfolio of connected use cases, with fewer but higher-value initiatives. Gartner predicts that by 2028 organisations that build AI portfolios will achieve twice the business value of those pursuing standalone use cases. Third, they invest nearly 1.8 times more in the data and knowledge-management foundations than in the AI use cases themselves.
Christopher Sladdin, Senior Director Analyst at Gartner, put it plainly: leaders must stop asking how to remove humans from interactions and instead ask where their involvement creates disproportionate value for customers.
Why this matters for SMBs
The survey covered large organisations, but the pattern is sharper in small and mid-sized companies. An SMB typically runs a single AI tool: a chatbot on the website or an auto-responder on email. If it is not connected to the CRM, the order system or the knowledge base, you get exactly the situation Gartner describes: an isolated use case, thin data and an impact nobody can measure.
The second lesson is measurement. The 4 percent figure is so low because most organisations never defined what “return” meant for them. Staff hours? Missed calls? Repeat customers? Without that number fixed up front, an AI project can never be a success; at best it “works”.
What to do in practice
- Fix one metric before you deploy. For example: what share of inbound calls the AI resolves without human intervention, or how many hours per week it frees up.
- Get the data foundation in order. Without an up-to-date FAQ, price list, delivery terms and customer history, even the best chatbot gets things wrong. This is where the organisations that do see a return spend the most.
- Connect the tools. A voice agent that can see the calendar and the CRM books appointments and records data. One that cannot only takes messages.
- Fewer use cases, but connected ones. Do not run five pilots in parallel; pick one process where the AI carries the case end to end, from intake to record.
- Analyse the conversations. AI’s biggest contribution is reading your calls and chats to show why customers keep contacting you, so you can remove that cause.
- Redesign the human role. Decide in which situations a customer must reach a colleague immediately, and build that into the system.
How we do it
The BoostYourBiz.ai AI Voice Agent and AI Chatbot follow the model Gartner describes in the organisations that do get a return: the solution is built on the systems you already run (calendar, CRM, knowledge base), every call and conversation is logged and analysed, and before go-live we agree together what we will measure. If you want to know where the measurable return sits in your customer service, book a consultation.
Source: Gartner, “Gartner Survey Reveals Just 4% of Customer Service Leaders Have Achieved the Expected ROI of Their AI Initiatives”, press release, 4 November 2026 (gartner.com)