At Verizon, AI Won't Fix Bad Incentives

This week PhoneArena published a conversation with an anonymous Verizon store employee describing how two internal AI systems, Personal Shopper and Priority Upgrades, shape daily sales work. The interview showed how automation enforces the same quota pressure that drives Verizon’s retail business. Personal Shopper is an AI tool built into the company’s point-of-sale software. It automatically fills shopping carts with phone upgrades, insurance plans, and streaming bundles. The employee said it often recommends irrelevant items, and if a rep doesn’t manually remove them, they remain in the final order. “It just builds a cart… pulls the most irrelevant items,” the worker said. “You could press three buttons and sell it to them.” The same report described strict sales quotas: at least 60 to 75 percent of all customers must leave with insurance or paid “perks” such as Netflix or Disney+. Missing those targets can lead to warnings or termination. A second system, Priority Upgrades, flags accounts likely to leave Verizon. The algorithm uses signals like transfer-PIN requests or browsing activity to identify customers at risk of churn. Employees said the system often mis-flags users, forcing manual fixes to avoid losing credit for the sale. Both tools serve the same purpos, that is, enforcement. They automate Verizon’s sales goals. Metrics define behavior Verizon presents its AI rollout as a customer-experience upgrade. In April 2025, the company told Reuters that a Google-powered assistant had cut call times and increased sales by nearly 40 percent. In June 2025, Verizon introduced a Gemini-based “Verizon Assistant” in the My Verizon app to handle billing and plan changes. Inside the stores, those same technologies operate differently. They push employees to sell more add-ons and discourage them from skipping optional products. “I don’t need an AI to tell me how to do it,” the employee said. “It’s the biggest nuisance I’ve ever seen introduced into a point-of-sale system.” Verizon’s public and internal narratives aren’t contradictory: they measure success in different ways. Executives track sales growth. Workers track pressure. Both reflect the same metric: revenue per interaction. Academic research on algorithmic management shows the same pattern in logistics, delivery, and call-center work. A 2025 study in Frontiers in Public Health found that algorithmic quota systems significantly increase stress and burnout among workers subject to automated performance tracking. A European joint research report on AI-based management systems concluded that such tools intensify work demands and reinforce existing managerial incentives rather than correct them. Together, the studies suggest that when algorithms are designed to enforce productivity metrics, they amplify pressure instead of improving judgment. Verizon’s AI tools fit that model. They perform exactly to the incentives they encode. A two-track AI strategy Verizon is expanding its AI operations on two fronts. At the enterprise level, it markets its network and edge infrastructure as a platform for AI workloads. The company partners with NVIDIA to host GPU-accelerated models on private 5G networks and with Amazon Web Services to link its network to Amazon Bedrock. These are large, technically sophisticated programs aimed at industrial clients. At the consumer level, the AI systems built into retail and customer support focus on retention and upselling. The first generates new business for corporate partners; the second extracts more value from existing subscribers. Competitors have taken narrower approaches. T-Mobile US uses an OpenAI-based assistant that helps support staff resolve issues but does not make direct product recommendations. AT&T applies AI mainly to network monitoring and predictive maintenance. Verizon stands out for tying its AI investments to an explicit sales-growth figure and using that figure as the measure of success. Automation meets intent AI doesn’t fix misaligned incentives. It scales them. Verizon’s retail AI systems automate the same sales logic that already governs its stores: maximize attachment rates, minimize churn, and push premium plans. The results are predictable. Employees face more pressure to meet targets. Customers face more upselling and confusion. And executives get the performance data the system was built to deliver.