Transfer success rates tell you the routing worked, not that the customer was helped
CX Today, in a Five9-sponsored piece published in October 2026, reports on Five9's 2026 Business Leaders Customer Experience Report. Per that report, 92 percent of organisations have implemented or piloted AI in customer experience. Yet 83 percent of consumers still repeat themselves when moving between AI and a human agent. Nearly all CX decision-makers believe their organisation preserves context. Five9's Sam Counterman says the industry lacks a clear definition of a successful handoff.
Why do handoffs fail when the dashboard says they worked?
Because most teams count transfers, durations and escalations rather than outcomes, so a handoff can be logged as successful while the customer still re-explains who they are and why they called.
The usual measures are transfer counts, handoff duration, escalation volume and the share of contacts resolved in self-service. They are useful, but none shows whether the problem was solved. Counterman calls it "a scoring gap, not necessarily a knowledge gap". For the customer, context survives only if the next person knows who they are, why they called, what they tried and what happens next. Routing to an agent without the right skills fails the same test.
How far should I trust the 83 percent figure?
Treat it as directional. It comes from Five9's own report, in a Five9-sponsored article, and the excerpt we read gives no sample size, method or definition of repeating oneself.
The excerpt also doesn't say whether 83 percent means every transfer or at least one. The comparison crosses two groups: consumers on one side, decision-makers on the other. Nearly all decision-makers claim context is preserved. Yet a sizeable minority admit customer repetition, limited agent visibility into earlier AI interactions and transfer glitches. The article doesn't say whether these are the same people. The 92 percent also lumps pilots in with live deployments, so a small trial counts the same as a full rollout.
What should I measure instead of transfer success?
Measure whether the customer reached an appropriate resolution and how much effort it took, across the whole journey rather than channel by channel, including repeat identification and re-explanation.
The article's example is a customer who goes from web chat to a virtual agent, then voice, then a specialist. Every leg can log as a clean transfer while the customer experiences one fragmented journey. That is why per-contact figures, such as the average handle time we have explained before, can look healthy throughout.
The coverage misses the plumbing. A transfer can succeed in routing while the history sits in a system the receiving agent never opens. Test that with your own calls.
What should I ask my vendor?
Ask what the receiving agent actually sees on arrival, how routing accounts for skills, and whether the platform reports resolution after a transfer rather than only that a transfer occurred.
Request a live demonstration of the agent's screen at the moment of transfer. Ask who defines resolution in the vendor's reporting. Ask how a resolved self-service contact is told apart from a customer who gave up. A supplier that can show only transfer counts has the scoring gap Counterman describes.
Frequently asked questions
What did Five9's report find about AI adoption?
That 92 percent of organisations have implemented or piloted AI use cases in customer experience, though adoption alone does not guarantee a better experience.
Is a completed transfer the same as a resolved contact?
No. Counterman's point is that moving a customer from AI to a human does not mean the problem has been solved.
Is the CX Today article independent?
Not in the usual sense. It is labelled as sponsored by Five9 and draws on Five9's report and a Five9 executive, so read it as one vendor's framing.
Sources: CX Today, article by Nicole Willing, reporting Five9's 2026 Business Leaders Customer Experience Report.