Turn Ping Wu into the
definitive voice on AI + CX.

Most companies don't know what their contact center actually sounds like to a customer. So we built a system that does. We called 63 enterprise contact centers—banks, insurers, telecoms, retailers. We navigated every IVR tree. We listened. ▪️ 71% were still running IVR menus with 4+ levels of nesting ▪️ 58% had no AI detection on any inbound channel ▪️ Only 12% offered a callback within the first menu level The average customer abandons after 2.5 minutes of IVR friction. That's not a technology problem. That's a prioritization problem.

Most VoC programs are built on data from the loudest 5%: the happiest and the angriest. What about the 95% in the middle who just quietly churned? We analyzed over a million contact center conversations. The pattern is consistent: ✅ Customers with one unresolved interaction were 3x more likely to churn in 90 days ✅ The signal was visible in the conversation — long silences, repeated questions ✅ CSAT surveys captured less than 8% of those at-risk customers The conversation IS the data.

I've been asked a hundred times: when will AI replace call center workers? Here's what I actually believe: The contact center is the central nervous system of your business. It's where customers tell you the truth — what they need, what's about to make them leave. AI doesn't replace that. AI amplifies it. ▪️ Co-pilots that surface the right answer before the agent searches ▪️ Voice agents that handle repetitive volume so humans focus on moments that matter Klarna tried to automate everything. They're hiring humans back. The organizations winning in CX right now are building unified platforms.

New CX leaders don't get a honeymoon period anymore. The board wants AI transformation. The CFO wants cost reduction. Both. Simultaneously. Here's the playbook that's working right now: ✅ Week 1-2: Don't touch the tech stack. Listen to 50 calls yourself. No summaries — actual calls. ✅ Week 3-4: Find the 3 highest-volume call types with consistent resolution paths. First automation candidates. ✅ Month 2: Run a quiet pilot. Measure AHT, CSAT, FCR — not just handle time. ✅ Month 3: Present results with real numbers. Not potential. Actual.

That last number is the one that matters most to me. It's easy to show a CEO a cost reduction. It's much harder to build something agents actually love using. When your agents love the tool, they use it right. When they use it right, customers feel it. When customers feel it, CSAT follows. The sequence matters. AI that's imposed on agents creates friction. AI that agents choose to use creates performance. We built Cresta to be the second kind.