CX leaders need authority over decisions

CX needs authority over operating decisions

A customer experience leader who spots a damaging launch needs a way to delay it. A larger team may improve the diagnosis while leaving the launch decision unchanged. A smaller standalone customer experience, or CX, function could have more influence if other departments must meet its requirements before proceeding.

The case for examining that authority starts with worsening customer outcomes. Forrester’s US Customer Experience Index declined for four consecutive years, reaching an all-time low in 2025, when 25% of US brands deteriorated and 7% improved. Discussing the year before, Forrester’s Rick Parrish described US consumer experiences as the worst in a decade; only 3% of companies qualified as customer-obsessed. Forrester sells research and advice on customer experience, so it has a commercial interest in companies treating these findings as a management priority.

These results follow two decades of increasingly sophisticated frameworks, maturity models, journey maps and certifications. Such tools help teams understand an experience, but authority determines whether they can change the decisions producing it. CX Day, held on the first Tuesday in October since 2013, gives the profession an annual occasion to explain its importance. Revenue and margin rarely need equivalent advocacy; there is no established Profit Day.

The practical question is what happens when CX requirements conflict with an operating target. Can an experience failure create a measurable cost for the responsible department? Can someone stop work already underway, and must a named person sign for the outcome? Those powers determine whether a concern raised before launch changes the launch decision.

Departmental incentives shape the customer’s experience

The need for those powers starts with how organizations divide work. In April 1968, Melvin Conway published in Datamation the observation now known as Conway’s Law: an organization’s communication structure shapes the systems it designs. Applied to CX, reporting lines shape coordination, budgets determine what gets funded, and incentives guide employees’ priorities. Customers encounter the combined result when they buy, seek help or request a return.

Retail’s separation of ecommerce and stores made that mechanism visible. Through the 2000s, most large retailers maintained separate ecommerce divisions with their own profit-and-loss accounts, inventory and buyers. A purchase could improve one division’s results while appearing to take revenue from another. The customer’s preferred way of shopping therefore created an internal dispute over sales credit.

Sales credit also affected individual pay, giving employees a reason to obstruct the intended experience. Some store associates hid popular merchandise to protect their commission earnings. CNBC described the conflict in 2014: “the single greatest obstacle to what the shopper wants is not technology, merchandise, or store operations. It’s the issue of revenue recognition.” Customer research could expose the damage, but the commission arrangement still rewarded the behavior causing it.

The same separation complicated buying decisions. As recently as 2016, Walmart’s digital and store buying teams lacked a formal way to coordinate their work. Macy’s had consolidated merchandising and marketing in 2015 into “one unified merchandising and marketing organization,” an approach contemporary analysts considered more aggressive than those of its peers. In these retail examples, removing separate profit-and-loss ownership resolved the competing departmental claims over sales.

CX leaders face a similar conflict when a small departmental budget comes with responsibility for an entire customer relationship. Improving that relationship requires decisions across service, product, pricing, marketing and operations, while the CX leader’s authority often ends at their reporting line. A dedicated CX team can also encourage operating departments to hand experience responsibility to that team. Surveys and quarterly indices then record the results after customers have passed through the processes that produced them.

A hypothetical contact center shows how this split in responsibility affects a launch. To meet a fourth-quarter cost target, it raises self-service deflection, the share of customers kept out of assisted service, from 45% to 60%, supporting a seven-figure reduction in cost to serve. The vice president, or VP, of Service owns the target and deadline, while CX asks whether customers have actually resolved their problems. If abandoned attempts count as successful deflections, both the financial case and the reported containment rate, the share of interactions that stay within self-service, improve while customers’ problems remain unresolved.

The launch proceeds because the quantified cost commitment carries more authority than the CX concern. Lower renewal rates reveal the experience cost later, after the business has accepted the apparent saving. A defined resolution measure and a right to pause deployment would bring that cost into the launch decision. Without those controls, Service directs the action and CX measures the consequences.

Forrester’s forecasts extend this concern to artificial intelligence, or AI, self-service. For 2026, it expected one-third of companies to damage experiences through frustrating AI self-service. It also forecast that 15% of CX teams would enter a reporting-driven “death spiral,” becoming replaceable measurement functions, while two-thirds would abandon journey mapping after repeatedly failing to change their organizations. These forecasts describe the risk of separating diagnostic work from decisions.

Make requirements binding through costs, intervention and ownership

The gap between diagnosis and action suggests a three-part test of authority. Does failure produce a measurable cost attributable to the responsible decision? Can someone stop work already in progress, and does a named individual approve the outcome? This proposed test examines what happens when teams disagree, because disagreement reveals whose requirements the business must satisfy.

The first part connects customer outcomes to resource decisions. A complaint can be thoroughly documented while remaining separate from the budget or performance target that encouraged the damaging action. Assigning its cost to the responsible operating decision makes the consequence relevant to performance reviews and future approvals. The decision owner then has to account for both the planned saving and the resulting harm.

Information security shows how consequences can gain that force. For roughly 15 years during the 1990s and early 2000s, security relied heavily on advocacy: an information-security director might sit three levels inside information technology, or IT, use awareness posters and training, and repeatedly lose budget requests. Public breach announcements made failures more visible. Industry bodies, governments and insurers then introduced obligations that changed the conditions under which companies could operate.

Those obligations connected security failures to commercial and legal consequences. California’s 2003 breach-disclosure development made breach consequences public, while credit card networks and card brands imposed and enforced merchant security requirements. Cyber insurance underwriters made specified controls a condition of coverage. Each mechanism reduced an operating team’s freedom to disregard a security requirement.

Executive reporting obligations strengthened that connection. In 2023, the Securities and Exchange Commission, or SEC, added disclosure of material incidents through an 8-K, a report of significant corporate events, and disclosure of board oversight in annual reporting. Security failures therefore affected formal reporting duties as well as technical repairs. Security approval increasingly became a condition that work had to satisfy before completion.

The second part of the authority test asks what happens when a requirement goes unmet. Microsoft’s Security Development Lifecycle, its process for incorporating security into development, makes the consequence explicit after a failed final security review: “the security advisor cannot approve the project, and the project cannot be released.” The review directly controls release approval. Giving a CX review equivalent authority would require the business to resolve a defined experience failure before proceeding.

Quality management offers another way to connect requirements with action during production. Edward Deming, described as the Godfather of Quality, argued in his third point for building quality into products and processes rather than relying on final inspection. Accounting for the financial cost of defects made failures visible. Giving employees authority to intervene let them address those failures as work happened.

Toyota’s andon cord, which allows employees to stop the manufacturing line when a problem appears, puts that authority at the point of work. Its relevance to CX is precise: the person encountering a defined failure can interrupt the process producing it. An escalation-only process leaves production running while managers consider a report. Stop-work authority makes correction a condition of continuing.

Contracts can impose a similar condition between companies. Stellantis states that IATF 16949 certification, an automotive quality-management standard, “is a required condition before any business relationship,” while Boeing includes AS9100 aerospace quality certification in purchase-order terms. These requirements affect whether business can begin or proceed under the agreed terms. Their force comes from the commercial obligation they create.

An experience standard could create an equivalent obligation if the parties agree on an outcome they can assess and enforce. Asking whether a customer contract includes a Net Promoter Score threshold, a measure of customers’ willingness to recommend a business, exposes the distance between an internal ambition and a contractual commitment. The next step is to choose an obligation suited to the relationship. A general promise of a good experience leaves the parties too much discretion over compliance.

The third part of the test assigns responsibility for judging that compliance. A named signatory makes clear who approves a specified outcome and on what basis. Without that assignment, several teams may review evidence while final ownership remains unclear. A documented approval also identifies who must respond when performance falls short or someone requests an exception.

Each part closes a different gap in authority. A measurable cost may become a routine expense if nobody can halt the damaging activity. A veto may be applied inconsistently if no individual owns the judgment, while a signature has limited force if the signer cannot influence the decision. Security and quality show ways to connect financial consequences, intervention rights and individual responsibility; bringing those mechanisms into CX requires decisions about the outcomes each control should protect.

Shopping agents connect operating details to product selection

Customer-deployed shopping agents give businesses another reason to make those operating decisions explicit. An agent choosing products for a customer can assess product-data accuracy, review counts, policy clarity, current availability and automated support. Weaknesses in that information can affect its assessment and therefore its selection. A problem previously captured in customer feedback can influence whether a product receives consideration.

The CX-framework contributor advancing this argument describes shopping agents as having no brand loyalty and software-mediated selection consequences as occurring weekly. As someone who helps develop CX frameworks, the contributor has a professional stake in how businesses organize and evaluate customer experience. The proposed mechanism focuses attention on whether policies and operating data are usable by the systems making selections. An unclear returns policy, for example, creates an interpretation problem that brand familiarity alone cannot resolve.

Research associated with Columbia and Yale shows which details can influence that interpretation. In real-world shopping tests, agents responded to price, ratings, review volume and a product’s position on a page. A “Sponsored” label reduced selection, while an “Overall Pick” endorsement increased it. Position bias persisted in text-only interfaces, allowing the researchers to rule out visual artifacts as the explanation for its persistence.

Selection also changed sharply when the underlying model changed. In those tests, one fitness watch’s selection share rose from 45% to 77% between two versions of the same AI model. Another watch fell from 25% to 6% across one model update. These results show why businesses need to distinguish changes in agent behavior from the measured effects of their own experience improvements.

Despite that variability, Adobe’s assessment of retail-page machine readability identifies information businesses can examine directly. Product pages were the lowest-scoring category, averaging 66%, with scores ranging from 54.2% to 82.5%. Adobe sells digital experience and commerce technology, giving it a commercial interest in investment in this work. Its findings point to a practical problem: information that people can navigate may still be difficult for software to interpret.

Adobe’s traffic findings explain why that problem has growing commercial relevance. In May 2026, AI-referred traffic to US retail sites increased 138% year over year and reached 14 times its October 2024 baseline. That traffic converted 54% better than non-AI traffic, reversing the underperformance Adobe described for the previous year. Traffic growth, conversion and page readability remain separate findings; measuring one does not establish the others.

The channel’s share of total traffic puts that growth in context. Bain, whose consulting business benefits from demand for advice on changing commercial channels, estimated that AI accounted for up to 25% of referral traffic at selected retailers but less than 1% of overall site traffic. The percentages use different denominators: one concerns referred visits, while the other concerns all visits. Businesses therefore have a growing reason to assign responsibility for agent-readable information while accounting for the channel’s current scale and changing selection behavior.

A smaller CX team needs enforceable responsibilities elsewhere

Growing commercial exposure makes organizational authority more important, but it does not determine the right size for a CX department. Human resources, or HR, shows how a function can spend decades debating its standing: Dave Ulrich questioned its continued existence in Harvard Business Review in 1998, and Ram Charan advocated splitting it in two in 2014. HR Executive later published “HR fought for a seat at the table. Now it’s slipping away.” The CX-framework contributor describes this as a twenty-eight-year recurrence of the same legitimacy debate.

Declarations of importance have offered little protection against that recurrence. “People are our greatest asset” has faced criticism as an empty slogan since at least 1984, illustrating the distance between declared value and operating authority. The contributor places CX closer to HR’s repeated advocacy than to security’s enforceable requirements. Marketing provides a partial qualification because its measurement and attribution have improved, although some organizations still struggle to secure recognition for the function.

Executive appointment trends can also reflect different organizational changes. Strategy& tracked new chief digital officer appointments falling from a peak of 160 in 2016 to 54 in 2018, concluding: “As transformation becomes part of the core business, the next step will be for the CDO to disappear.” PwC states that “the role of a good CDO is to put themselves out of a job.” Strategy& and PwC have commercial interests in advising companies on transformation and organizational design; both statements describe success as transferring responsibility into everyday operations.

Customer-related appointment figures raise the same question about what changes beneath the title. In Marketing Week’s 2026 survey, 1.3% of respondents had gained a new chief customer officer during the past year. Among business-to-consumer, or B2C, respondents, 3.6% had gained a chief growth officer. Because the respondent groups differ, these percentages cannot be compared directly, and the appointment rates alone do not describe the authority held by operating teams.

The contributor suggests that a disappearing title and a successfully integrated discipline can look similar externally for about three years. Quality offers a way to distinguish them: a smaller inspection department can coexist with requirements embedded in production, contracts and employee decisions. A smaller CX function could follow that pattern if its responsibilities become binding elsewhere. Departmental reduction would then accompany a transfer of authority into operating work.

Before making that reduction, executives need to identify who accepts each responsibility and where the obligation is recorded. A team that inherits a customer outcome also needs the authority to govern the decisions producing it. Toyota’s stop-line example provides a concrete test: can someone encountering a defined failure interrupt the process? Budget and reporting-line changes should follow an examination of which controls will remain in force.

Negotiate a launch veto, a contractual standard and an accountable signature

Turning that examination into a planning decision requires a clearly defined request for authority. Experience responsibility can extend across operating teams, as intelligence and operations do, while a central group maintains requirements those teams must meet. The CX-framework contributor presents Marketing Operations as making a similar shift from support work toward coordination across functions. Managing shared infrastructure can provide a route to approval authority because the responsible function already participates in completing the work.

Before seeking new powers, CX leaders should examine the incentives and external obligations already shaping that work. The retail cases make sales credit, commissions and departmental financial targets a useful starting point: identify where they reward behavior that harms the intended experience. Journey mapping can then describe a process whose underlying conflicts are understood. Reviewing existing costs and rules can also reveal where an experience approval requirement would have practical force.

The first request is a veto over one critical decision type, such as deploying customer-facing AI. The agreement should specify the outcome that permits launch and the failure that requires a pause. For self-service, it should assess maintained resolution rates alongside containment so abandoned attempts cannot count as evidence of success. This would give the hypothetical contact-center launch its missing control: a defined decision right.

Exercising that veto requires an agreed process for assessing and acting on evidence. The business must identify who measures resolution, who can pause deployment and what must change before approval resumes. An attributable commercial consequence also connects the experience outcome to the operating owner’s commitments. The review then becomes part of the release decision.

The second request is an experience standard written into one contract. It could sit in a supplier service-level agreement, which specifies expected service performance, a partner agreement or a platform vendor’s agent-readiness requirement. The parties must agree on an assessable obligation and the consequence of failing to meet it. The resulting commitment gives a specified experience outcome contractual force.

An operating audit can help select that obligation. Product-data accuracy, returns-policy clarity and product-page machine readability are concrete candidates because they affect how customers and shopping agents assess an offer. The audit should identify which team or partner controls each weakness before responsibility enters a contract. Changing model preferences make this assignment important: a business can govern the quality of its information more directly than an agent’s selections.

The third request is documented approval from one named executive for a specific experience outcome. The chief financial officer’s, or CFO’s, explicit sign-off provides a model for recording responsibility; the executive chosen for CX needs authority over the relevant operating decisions. The agreement must define the evidence required for approval, the executive’s power to require corrective action and how exceptions will be handled. Those terms make the signature an exercise of authority over the outcome the executive accepts.

Key takeaways for decision-makers

• Authority determines CX impact: CEOs can give CX leaders defined approval rights over decisions that affect customer outcomes. Department size offers little indication of whether the function can change a damaging launch.
• Incentives shape customer outcomes: Operating owners need to examine where commissions, departmental budgets and cost targets reward harmful behavior. Self-service targets, for example, need resolution measures that distinguish successful support from abandoned attempts.
• Binding controls turn findings into action: CX authority needs measurable consequences, intervention rights and a named approver. Security and quality provide models for making requirements enforceable through release gates, contracts and stop-work powers.
• Shopping agents increase exposure to information failures: Commerce and product teams need clear ownership of product-data accuracy, policy clarity and machine readability. Agent selection also changes with model updates, so selection shifts require investigation before being attributed to experience improvements.
• Smaller CX teams require explicit transfers of responsibility: Before reducing a standalone function, executives need to identify which operating teams inherit its obligations and decision rights. Embedded requirements preserve influence when reporting lines and budgets change.
• Start with specific approval rights: CX leaders can negotiate a veto over one critical launch type, an assessable standard in one contract and documented approval from a named executive. Each agreement needs evidence requirements, corrective actions and exception rules.