AI Support Resolution Gap: Setup, Not Software, Drives Results

What happened

Wonderchat, an AI customer-support vendor, has published a study of 5.85 million support conversations showing a 42-point gap in resolution rates between businesses running the same underlying AI. Coverage of the research by CustomerThink reports that the difference came down to how each business had set up its deployment, not to the technology itself.

According to the reporting, the research points to three factors that separated effective AI support from underperforming implementations: how completely the knowledge base is documented, whether the AI has access to the systems it needs to act, and how clearly escalation rules define when a conversation should pass to a human.

The finding is significant because it reframes a common assumption in AI procurement, namely that platform choice is the main determinant of performance. Based on the summary available, the study suggests that configuration and operating design account for a large share of outcomes.

Why it matters

For leaders in experience and digital transformation, the study shifts attention from which model to buy toward how the model is built into the service operation. An AI agent can only resolve what it can read and act on. Gaps in documentation mean the agent cannot answer, gaps in system access mean it cannot complete the task, and unclear escalation rules mean customers either get stuck in loops or are handed off too late. Each of these is an operational decision, not a software feature.

The implication extends beyond customer support. Any organisation deploying AI into a service journey, whether in government, financial services or telecoms, is effectively redesigning that journey. The study suggests that the quality of the underlying knowledge, integrations and handover logic will determine whether automation feels helpful or frustrating to the person on the other end.

By the numbers

• 5.85 million support conversations analysed in the Wonderchat study.
• 42-point resolution gap between businesses running identical AI.
• Three factors identified as the key differentiators: documentation coverage, system access and escalation rules.

The Renascence take

Most coverage of AI support will focus on the headline gap. The more useful question is why identical tools diverge so sharply, and the answer is that most organisations treat deployment as a switch-on exercise rather than a service design project. The behavioural principle at work is simple: customers judge an interaction by whether it moves them forward, so every missing answer or dead-end integration registers as a broken promise.

The 42-point gap is not a technology story; it is a design story. Organisations that treat their knowledge base, system permissions and escalation paths as core customer experience assets, rather than IT housekeeping, will outperform peers using the same models. A customer-obsessed operator should map the top failure journeys first, audit whether each answer exists in writing, and set escalation triggers based on customer effort rather than on what is technically easy to hand off.