The Asymmetric Service Economy: Why Customer Support Is Broken on Both Sides

Customer service has a paradox at its center. It is the function inside large companies where investment goes up year after year: in headcount, in tooling, in training, in software, while the outcomes those investments are nominally meant to produce go in the opposite direction. Net Promoter Scores for service-heavy industries trend down across the decade. Hold times go up. First-contact resolution rates stagnate. Repeat-call rates climb. And the consumers experiencing this decline are not imagining it: independent measurement of the customer-service experience, including the long-running National Customer Rage Study, has documented year-over-year deterioration in a remarkably consistent direction.

The paradox is not that service is bad. The paradox is that no amount of incremental investment from inside the existing model meaningfully improves it. We argue in this paper that the reason is structural: not effort, not technology, not management. The shape of the relationship between consumers and the brands they pay is asymmetric by design, and only an architectural intervention can fix it.

The brand-side compound

Look at the customer-service operating model from inside a Fortune 500 retailer or telecom. Three forces compound on the brand side, each making the system measurably worse over time.

Per-contact cost rises with no offsetting gain. Wage inflation, training time, compliance overhead, and tooling sprawl mean the marginal cost of resolving a single customer interaction has increased every year of the past decade. Per-contact outcomes have not. The business case for "a better resolution" never wins the quarter; the business case for "fewer contacts" always does.

Deflection metrics distort what the system optimizes for. Modern customer-service tooling is overwhelmingly oriented toward "ticket avoidance": chatbots, IVRs, gated help pages, self-service portals, and the budgets for those tools are justified by deflection rates. But deflection is cost-shifting, not resolution. An interaction that gets deflected often resurfaces later as a churn event, a chargeback, a regulatory filing, a social-media complaint, or a litigation threat. The original cost is moved downstream where it is harder to measure and more expensive to fix. Brands optimize for the deflection metric they can see and accumulate the consequences of the resolution they couldn't.

Personalization at scale is operationally impossible. The interactions that actually matter: retention saves, dispute judgment, benefit clarifications, upsell at the right moment, require per-customer reasoning that human agents are too expensive to scale to and current chatbots cannot reach. The conversations where the company has the most to gain are the ones least served by either economic option. The margin is left in the queue.

The compound effect is that customer service is roughly a half-trillion-dollar global line item that consistently produces the lowest customer satisfaction scores in any given company. Brands know this. They lack the technology to fix it because the fix is not a better chatbot; it is an architectural change in who is sitting on each side of the conversation.

The consumer-side compound

The consumer experience of the same system is not the mirror image of the brand experience. It is its own distinct compounding failure, and the central observation is uncomfortable: the consumer experience has been deliberately engineered to be hostile, because hostility is profitable.

The IVR maze. The average billing or dispute call routes through five or more menu layers, seven to fifteen minutes on hold, and at least one transfer. Voice menus are not optimized for resolution. They are optimized for deflection, every branch funnels the caller away from a human with authority. Most consumers abandon before they reach one. Abandonment is, by the deflection-metric logic, a success.

No front door. Getting in touch with a company has become a scavenger hunt. Email is deprecated. Live chat is hidden behind login walls. Support URLs 404. Social-media channels are performative. Phone numbers rotate between 1-800 trees. A determined consumer often cannot find the right queue for their problem, and that is before anyone has said no.

Engineered friction. Cancellation requires calls where signup required clicks. Refunds require forms that require account numbers that require earlier paperwork. Retention scripts are written to exhaust the caller; deflection UX is A/B tested against resolution. These are not bugs. They are architectures that convert consumer time and fatigue into retained revenue. The math works for the brand: the average household leaves real dollars on the table every year, somewhere between $1,200 and $2,500 in unnegotiated bills, unclaimed entitlements, and undisputed charges. Not because the money is unavailable, but because the path to "someone who can decide" has been fortified against them.

Asymmetry of access is the product. This is the load-bearing observation. The consumer has an evening and a hold queue. The brand has dedicated retention teams, pricing engines, legal-review processes, and software designed to wear down the determined caller. The asymmetry is the business model. It is what allows the brand to win the negotiation that is theoretically taking place between two parties.

Why incremental fixes fail

Both sides of this system have spent the last decade trying to fix it from inside. Brands have invested billions in CX software. Consumers have organized in Reddit threads, used apps like Rocket Money to surface unwanted subscriptions, and adopted bill-negotiation services. Both sides have produced incremental improvements. Neither has produced structural ones.

The reason is that the fundamental asymmetry remains in place. A better chatbot does not change the fact that the consumer is a single human with limited time facing an institution that has dedicated infrastructure for the conversation. A better bill-negotiation app does not change the fact that the consumer is still operating asynchronously through a friction layer the brand controls. The asymmetry of compute, attention, and authority is the load-bearing flaw. Until both sides have agents that are competent, persistent, and accountable, and a protocol to transact between them, the system will continue to underperform on its own metrics.

The structural fix is symmetrical

The intervention that actually changes the shape of the problem is to give the consumer something the brand has had for thirty years: an agent. Not a chatbot. An advocate that is competent, persistent, accountable to the consumer, and capable of transacting on the consumer's behalf with the brand's agent under bounded authority.

Symmetric agency reframes every interaction. The consumer advocate does not wait on hold. The brand agent does not hide behind a five-layer IVR. The conversation actually happens. Both sides have the compute and the authority to settle the dispute, negotiate the bill, claim the benefit, or process the cancellation in a single round-trip. The receipt is signed, audited, and inspectable by the human on either end.

This is the architectural insight that the agent-to-agent commerce category is being built around. It is not "AI for customer service." It is the elimination of the asymmetry that has made customer service economically broken for both parties.

Implications for incumbents

Three implications follow.

First, brands that prepare for the symmetric model will outperform brands that resist it. The asymmetric model is profitable in the short run for the brand and expensive in the long run for the brand's customer-lifetime value. Brands that adopt agent-to-agent commerce early will capture the customers whose dissatisfaction is currently absorbed by the system rather than acted on. The first-mover advantage in symmetric service is real, and it accrues to the brand, not just the consumer.

Second, the regulatory environment is shifting underneath the asymmetric model. Click-to-cancel rules, dark-patterns enforcement, and data-portability mandates are all making engineered friction more expensive to operate. Brands that have built their margins on the asymmetry will face compounding regulatory costs over the next five years. Symmetric service is not just an upside; it is a hedge against the regulatory direction.

Third, measurement will shift. Today's CX metrics: handle time, deflection rate, first-contact resolution, are designed to measure the asymmetric system. They will be replaced over the next decade by metrics that measure outcomes: time-to-resolution-without-human-effort, percentage of in-flight disputes settled by agent-to-agent, customer-financial-recovery per active advocate. Brands that adopt the new metrics early will be operating in a different feedback loop than brands that don't.

Customer service is not broken because the people running it are bad at their jobs. It is broken because the architecture is asymmetric, and incremental investment cannot fix architectural problems. The structural fix is to give both sides an agent. We expect that fix to define the next decade of customer engagement.