Why deflection is the wrong metric for AI in customer success

Why deflection is the wrong metric for AI in customer success

Deploy AI to solve one problem the wrong way and it can create more problems for your business. Rebecca Holland, [https://www.linkedin.com/in/rebecca-d-holland/] Founder at GTM RH Advisory and VP of Global Customer Success at Vimeo [https://vimeo.com/product-updates], has worked through five acquisitions. She explains what always breaks in customer success during a merger, and why customers do not care that you just got acquired. She makes the case for customer success as a revenue engine, why deflection makes her stomach turn, and how repurposing people the way IKEA did turns automation into new revenue. Let AI handle the slog, she says, but never outsource your brain. Topics discussed:  * Why deflection is the wrong customer success metric * What always breaks in customer success during acquisitions * Treating customer success as a revenue engine, not cost * Using AI to unify disparate customer data systems * The three tests every AI tool must pass * Repurposing specialized people into new revenue, like IKEA * Writing playbooks for 80%, leaving 20% for nuance * Resourcing AI adoption with an internal enablement specialist

Why outages might actually be your biggest CX opportunity

Drew Candres [https://www.linkedin.com/in/andrewjcandres/] built support organizations where a single tweet on a Saturday night could move a market 10% and flood your queues while the rest of the world slept. Now as VP of Customer Experience at GameChanger, he's applying those same instincts to a completely different emotional context: parents trying to capture their kids' first home run. The contrast makes for one of the more grounded and honest conversations about what it actually takes to build CX operations that hold up when things go sideways. One of the sharpest observations in this episode: some of his highest CSAT scores, across every company he's worked for, came on outage days. When handled well, those days produced scores around 50% above the average. His argument is that customers already know things break. How you handle it is the only thing they're actually judging. Topics Discussed: * Why containment rate is a meaningless metric and what it actually measures * Manual support as the most expensive technical debt a scaling startup can carry * The two-step framework for deciding what to automate and what to fix first * Rate of change in inbound volume as an early warning system no platform has fully solved * Why data alone doesn't move product teams, and the storytelling structure that does * Embedding proactive support into the build process through phased rollouts and power user cohorts * Why CX foundation work must come before executive headcount, not after * Why emotional connection becomes the primary differentiator as AI compresses product advantages

Why CS Playbooks Are Failing Your Team

Most CS leaders respond to scale pressure by automating SMB and protecting enterprise. Leana Hart [https://www.linkedin.com/in/leanahart/], Director of Customer Success at Axon [https://www.axon.com/careers], thinks that binary is costing companies on both ends — and she's spent her career managing the full spectrum, from SMB through Fortune 500 enterprise, to prove it. In this episode, Leana gets specific: the account tiering logic she actually uses, why she thinks the playbook-first approach backfires, and the question she teaches her team to ask customers that surfaces renewal risk faster than any health score. Topics discussed: * Why ARR-based segmentation misses influential smaller accounts and underweights expansion potential in the SMB tier * The 4-signal churn indicator Lena uses to identify at-risk accounts within a high-volume book: support ticket volume, webinar attendance, call recency, and 90-day product adoption trend * Why rigid playbooks create noise rather than behavior, and what she builds instead to get CSMs thinking critically * How quarterly portfolio reviews convert individual CSM wins into trackable revenue impact ahead of QBRs and annual reviews * The direct renewal-readiness question she asks customers mid-cycle, and how to read the hesitation in that answer before it becomes a lost deal * Why SMB retention is structurally worse than enterprise, and the specific investment gap on both sides that drives it * Where tools like Claude and Cursor are already changing what CSMs can do with customer data, and the skill gap preventing most teams from getting there * Why project management is the most under-hired and under-trained skill in CS, especially as enterprise account complexity grows Listen to more episodes:  Apple [https://podcasts.apple.com/us/podcast/ai-cx-innovators/id1802013873]  Spotify [https://open.spotify.com/show/0XpZj9bW7OkgXYgRX4DOTJ?si=14589b77ec1d4d00]  YouTube [https://www.youtube.com/playlist?list=PL6T5PUa9NwVKi275ahocKKL1ZG341d2-u]

How OpenAI won the AI race through experience design, not the model

Vishal Anam [https://www.linkedin.com/in/vishalanam/] has spent 20 years helping global organizations close the gap between measuring customer experience and actually acting on it. As Head of CX Consulting, Research and Analytics at Datamatics [https://www.datamatics.com/get-in-touch/sales-enquiry], he has worked across enough enterprise programs to know exactly where they break down. Most CX programs do not fail because organizations stop caring. They fail because the infrastructure for decision-making never gets built alongside the infrastructure for listening. Vishal makes that case clearly here, and backs it up with specific examples drawn from client work, including a contrarian read on why OpenAI captured early market dominance that had nothing to do with their underlying model. Topics discussed: * Why NPS without decision pairing produces no real change * The cognitive load framework for reducing CX decision fatigue * How macro and cultural factors silently distort NPS scores across regions * The SOP gap that breaks CX programs at scale * Optimization vs. transformation: how to tell which one you actually need * Why OpenAI's early lead came from onboarding design, not model quality * Building the CFO business case around missed revenue, not satisfaction scores * How AI is shifting the analytics function away from normalization toward action * Why promoter neglect is a bigger missed opportunity than most teams realize Listen to more episodes:  Apple [https://podcasts.apple.com/us/podcast/ai-cx-innovators/id1802013873]  Spotify [https://open.spotify.com/show/0XpZj9bW7OkgXYgRX4DOTJ?si=14589b77ec1d4d00]  YouTube [https://www.youtube.com/playlist?list=PL6T5PUa9NwVKi275ahocKKL1ZG341d2-u]

Means, motive, and opportunity: why LendingTree uses a crime framework to train CX empathy

Brock Thompson spent 22 years in digital insurance and financial services, starting as a call center agent and eventually becoming VP of Customer Fulfillment at LendingTree, where he oversees inbound and outbound calls, SMS programs, calls technology, and the company's AI voice division. Having sat on both sides of the phone shapes how he thinks about where technology belongs in high-stakes consumer conversations. In this episode, he gets specific about how LendingTree built and justified their first consumer-facing AI voice product, how they brought skeptical call teams along, and why their approach to AI adoption looks more like a marathon than a sprint. Topics Discussed: * The three-part "stickiness" framework: emotional, technological, and financial * Starting the AI business case with one linear, fully measurable conversation path * "Gut drives the test, data drives the decision" as a buying and build philosophy * Why legal and compliance are the first call, not the last, in any AI rollout * The means, motive, and opportunity framework for training human agent empathy * Where human agents still outperform AI in financial services conversations * Reframing AI to call teams as reallocation of effort, not elimination of roles * Evaluating vendors by whether they understand your actual problem, not just their product * The rule of one: prove it once, measure it once, then scale Listen to more episodes:  Apple [https://podcasts.apple.com/us/podcast/ai-cx-innovators/id1802013873]  Spotify [https://open.spotify.com/show/0XpZj9bW7OkgXYgRX4DOTJ?si=14589b77ec1d4d00]  YouTube [https://www.youtube.com/playlist?list=PL6T5PUa9NwVKi275ahocKKL1ZG341d2-u]