Contact center leaders are declaring victory on artificial intelligence. The financials tell a different story.
A new survey from AI platform provider Laivly found that 65% of customer experience leaders classified their most recent AI project as a success. Yet 43% of all AI projects are currently delayed or stalled, 53% have exceeded budget, and 28% of leaders say AI that cannot handle customer complexity has cost them revenue. Another 20% acknowledge they are losing money but cannot quantify how much.
The disconnect matters for collection operations, where contact center AI is being deployed at record speed under intense pressure from above. The survey of 200 contact center leaders, conducted in March, found that 43% of boards are dissatisfied with AI progress, and that impatience is pushing organizations to launch tools before they are operationally ready.
The revenue stakes are significant. Nearly half of companies, 49%, report increased customer friction directly tied to their AI tools, and 57% of those companies say they are losing between 5% and 10% of sales as a result. Companies that reduced friction saw the opposite: 36% report AI actively growing revenue by 5%.
The report also challenges a core assumption behind many AI business cases. While 78% of companies expect savings through agent headcount reduction, and 44% plan cuts within 12 months, Laivly found the most aggressive cutters are the same companies reporting higher friction, greater revenue leakage, and higher project costs.
Frontline resistance is another drag. While 31% of leaders cite fear of job replacement as a reason agents avoid AI tools, the bigger issue is reliability. Forty-six percent say agents struggle with AI that lacks context across interactions, and 36% say it introduces compliance and tone risk, a finding that should resonate in an industry where a single noncompliant statement can trigger litigation. Untrustworthy tools have consequences: 36% of companies saw agent turnover rise last year.
Tool sprawl compounds the problem. More than half of companies, 56%, run more than three AI tools, and most of those report increased customer friction. Unified stacks are four times more likely to stay on budget.
What separates the winners is not more AI but better execution. Half of successful deployments prioritized use cases with measurable ROI in the first 90 days, and 61% of companies that invested in real-time agent guidance and coaching expect revenue growth from faster, more accurate service. Looking ahead, 54% are preparing for agentic AI that completes actual transactions such as refunds and account updates.
