Back
EP 002
34 Min
Zack Hamilton has advised more than 200 retail brands and watched companies post NPS scores in the 70s and 80s while their same-store revenue fell and first-time customers lapsed at 70%. His verdict: the score is not evil, it is dangerously incomplete.
Zack Hamilton looked at the customer experience industry and decided it was measuring the wrong things, reporting to the wrong people, and wondering why no one cared. As the creator of the Experience Performance System and host of the show Unf***ing Your CX, he argues that CX has to become a performance function, not a reporting one. On this episode of the Fini Podcast, he made the case against NPS, walked through the business case that actually wins budget, and explained why most AI support deployments stop at containment.
Meet Zack Hamilton
Zack spent over two decades advising more than 200 retail brands, including Macy's, Sephora, Home Depot, and Coach, and served as Chief Experience Officer at Aaron's. He created the Experience Performance System, a framework Fortune 500 companies use to turn customer insight into business performance, and he hosts one of the more provocative voices in the field for CX leaders tired of being treated like a call center.
Why CX programs really get defunded
Zack's first hard truth: CX programs do not get defunded because they fail, they get defunded because they become irrelevant. Practitioners get obsessed with slicing data and building dashboards, and slowly disconnect from execution. Executives do not want more dashboards. They want faster, better decisions, less risk, more growth, and less churn. If CX is not changing operational behavior, funding becomes optional, and optional programs get cut.
NPS is dangerously incomplete
Zack does not hate NPS, he calls it dangerously incomplete. It measures advocacy and sentiment after the fact, not behavior. Brands report the score and close the inner loop, but skip the hard outer loop where transformation happens, so the metric gets gamified. The danger is false confidence: he has seen companies with NPS in the 80s whose P&L is bleeding, with first-time digital customers lapsing at 70 to 80%. Meanwhile friction compounds and silent churn accelerates while the score says everything is fine. What he watches instead is decision velocity, time to identify friction, customer lifetime value, and signal health.
The one-page business case
The tool Zack teaches is a one-page business case, not a dashboard. He learned it the hard way at Aaron's, when he showed up to a prioritization session with NPS trends and customer stories and was told, good data, keep monitoring it. His cross-functional partners, by contrast, brought specific business cases that drove decisions. So he rebuilt his around a real friction: customers abandoning carts because of delivery-date uncertainty. He translated it into a problem statement with a time period, the number of customers affected, and the financial impact, then added the cost of inaction, a recommended approach, and the exact resources required. Co-created with the partner who owned the metric, it changed the conversation from sentiment to business impact. As he puts it, executives do not fund feelings, they fund returns.
AI as containment vs AI as a business system
Zack's sharpest AI point is that most teams treat it as a containment tool instead of a business system. Agents close 80 to 90% of tickets, satisfaction looks fine, and you cut human cost, so on the surface it looks great. But you have only moved the cost to serve from human to agent. That is automation of mediocrity, not transformation, because you keep resolving the same tickets instead of fixing the upstream friction that generates them. His model: AI should detect the friction, quantify the impact, recommend and complete the action, then trigger a workflow to fix the root cause. The agent is a direct reflection of the human-led system behind it.
What support leaders should take from this
Be a performance function, not a reporting one. If your work does not change decisions, priorities, or resource allocation, it is a marketing program for feelings.
Translate friction into a business problem. Name the time period, the customers affected, the financial cost, and the cost of inaction.
Do not trust a high NPS. A great score can sit on top of a bleeding P&L. Watch behavior, friction, and decision velocity.
Bring a one-page business case. Co-create it with the cross-functional partner who owns the metric, and specify the exact resources.
Use AI to fix friction, not just contain it. Closing the same tickets forever is mediocrity at scale. Feed resolutions back into fixing the upstream cause.
Earn buy-in with receipts. Nobody hands you a seat at the table. You earn it through friction identified, cases built, and results delivered.
Listen to the full episode
Zack goes deeper on decision velocity, agentic CX, and why the power hierarchy inside companies is the real barrier, in the full episode of the Fini Podcast. You can find him on LinkedIn and on his show, Unf***ing Your CX.
AI that detects friction and fixes it upstream, not just contains tickets, is what Fini is built for. Book a demo to see it on your own data.
Why does Zack Hamilton say NPS is lying to CX leaders?
He calls NPS dangerously incomplete rather than evil. It measures advocacy and sentiment after the fact, not behavior, and teams report the score while skipping the hard outer loop. The result is false confidence: companies post NPS in the 80s while revenue falls and customers lapse, so the score hides compounding friction and silent churn.
What metrics should CX leaders use instead of NPS?
Zack tracks decision velocity (how fast you go from identifying friction to acting on it), time to identify friction, customer lifetime value, and signal health across surveys, behavior, and operational data. The aim is a balance of financial outcomes and the performance of your experience system.
What is a one-page business case in CX?
It reframes a customer friction as a business problem: the time period, number of customers affected, and financial impact, plus the cost of inaction, a recommended approach, and the exact resources required. Co-created with the cross-functional partner who owns the metric, it drives prioritization because executives fund returns, not feelings.
What is the difference between AI as containment and AI as a business system?
Containment closes tickets and cuts cost but keeps resolving the same issues, which Zack calls automation of mediocrity. AI as a business system detects the friction, quantifies its impact, completes the action, and triggers a workflow to fix the upstream root cause so fewer tickets arrive in the first place.
