There is a number that looks great in a board presentation and actively misleads the people running it: deflection rate. For years, contact centre leaders have been sold AI on the promise of keeping customers away from human agents. The fewer calls that reach a person, the better the ROI story. Zoom's latest deep-dive into customer experience management challenges that logic directly — and the argument deserves serious attention from anyone responsible for a customer-facing operation.
The Deflection Trap
Zoom's research makes a pointed distinction between AI that deflects and AI that resolves. Deflection means a customer did not reach an agent. Resolution means a customer got what they needed. These are not the same thing, and conflating them is one of the most expensive mistakes a CX operation can make. A customer who abandons a chatbot after three failed attempts has been deflected. They have not been served. Worse, they are now frustrated, and that frustration lands somewhere — in churn data, in review scores, in a social post you cannot take back.
The report frames the contact centre not as a cost centre to be minimised, but as a strategic asset. That reframe matters because it changes what you optimise for. If the contact centre is a cost centre, you optimise for volume reduction. If it is a strategic asset, you optimise for outcome quality — resolution rate, customer effort score, first-contact resolution, lifetime value retention. AI looks very different when measured against those criteria.
What Good AI Actually Does in a Contact Centre
The AI applications that genuinely move CX metrics are not the ones replacing agents wholesale. They are the ones making agents faster, better-informed, and less likely to make avoidable errors. Think real-time knowledge surfacing that gives an agent the right answer before the customer finishes asking the question. Think sentiment analysis that flags a conversation turning negative so a supervisor can intervene before the call escalates. Think post-interaction summarisation that eliminates the two minutes of after-call work that quietly consumes ten percent of agent capacity every shift.
These are augmentation plays, not replacement plays. And they require something that pure automation cannot provide: a human in the loop who can read context, exercise judgment, and adapt when the situation does not fit the script. That combination — AI handling the predictable, humans handling the complex — is where contact centre performance actually improves.
The Hybrid Model Is Not a Compromise
Operations leaders sometimes treat the hybrid human-plus-AI model as a transitional state — something you do until the AI gets good enough to go fully autonomous. Zoom's findings suggest that framing is wrong. The contact centre of the near future is not a fully automated one. It is one where AI and human talent are deliberately designed to complement each other, with routing logic, escalation paths, and quality frameworks built around that partnership rather than despite it.
This is precisely the model Conveneo is built around. Our multilingual agent teams work alongside AI tooling — not as a fallback when automation fails, but as an intentional design choice. Automation handles the high-volume, low-complexity interactions efficiently. Trained human agents handle the nuanced, emotionally loaded, or operationally complex ones. The result is a service operation that is both scalable and genuinely capable of delivering the kind of resolution experience that builds customer loyalty.
The Operational Takeaway
If your current AI investment is being measured primarily by how many contacts it deflects, it is worth asking a harder question: how many of those deflections were actually resolutions? Audit the drop-off points in your automated flows. Look at what happens after a customer abandons a bot interaction. Track whether deflected contacts resurface as repeat contacts, complaints, or silent churn.
The contact centre is not a problem to be automated away. It is a relationship channel that happens to have a significant operational cost. The smartest response to that tension is not less human involvement — it is better-designed collaboration between human expertise and AI capability. That is not a compromise. That is the competitive advantage.