The industry optimised for cost and quietly got worse at accountability. The companies that tie experience to revenue, margin and trust are about to separate from everyone still selling seats.
Here is an uncomfortable truth about the industry we work in, ourselves included: we became excellent at measuring effort and almost completely unable to measure worth.
We can tell you, to two decimal places, how many contacts were handled, how fast they were answered, and what customers scored us on a survey they barely read. We have dashboards for everything except the only question the board is actually asking: what did any of this do to the bottom line?
That gap is not a reporting inconvenience. It is the single largest unclaimed growth opportunity in most enterprises — and the reason customer experience, despite a decade of investment, still sits on the cost side of the ledger when it actually and factfully belongs on the growth side.
The opportunity hiding in plain sight
The evidence that quality customer experience moves money is no longer in dispute. McKinsey's research puts the prize plainly: improving customer experience for customers (not only for our companies conveniences) lifts sales revenue by roughly 2 to 7 percent, profitability by 1 to 2 percent, and total shareholder returns by 7 to 10 percent. Between 2016 and 2021, CX leaders grew revenue at more than twice the rate of laggards. In banking specifically, leading brands in customer experience delivered 55 percent higher total shareholder returns than low-performers over a decade.
And the cost of getting it wrong compounds quietly. McKinsey estimates that compensating for a single lost customer can require acquiring three new ones.
So the financial case is settled. The puzzle is why so few organisations have captured it. The answer is structural — and it has four parts.
Why the value leaks out
We measure with tunnel vision. The metrics that run most CX functions were never built to speak to a CFO. Net Promoter Score and CSAT tell you how a customer felt about a survey; they do not tell you what happened to revenue, retention or trust. Gartner has been blunt about it, predicting that more than three-quarters of organisations would stop using NPS as a measure of success in service and support, and titling its own research, memorably, around the idea that the metric is everywhere and useless where it's measured. McKinsey's CX leadership reaches the same conclusion from the value side: companies consistently fail to link experience to financial outcomes, and surveys do not solve the problem (McKinsey).
When the instrument can't see the outcome, the outcome doesn't get managed, nor rewarded.No one owns the result. In most organisations, strategy sits with one team, technology with another, daily operations with a third, and the insight function somewhere off to the side — often across several vendors. Each does its part competently. The customer still falls through the gaps and cracks between them. While the customer very easily recognizes that “she is missed”, the experience fails, and technically it is nobody's fault. A customer who can't complete a loan application doesn't care which box it belonged to. They just had a bad experience with your brand — and you may not learn why for months.
The intelligence is missing. The most valuable signals a business owns — the early warning of churn, the unspoken upsell, the fraud flag, the moment a frustrated customer decides to leave — are generated in interactions and then lost, either to cold automation that can't interpret them or to operations with no mechanism to act on them in real time. The value of an experience lives in the exceptions, and most systems are built to handle the average.
The architecture is patched, not engineered. CX capability has typically been bought in pieces — a platform here, an outsourcer there, an analytics tool later — at different times, on different logic. The result is an estate that no single party designed and no single party can be held responsible for. Fragmentation isn't just inefficient. It's why accountability has nowhere to land.
The false choice the market keeps selling
Faced with this, the market offers two tidy answers, and both are traps.
The first is people only: judgment and care, but a cost base that climbs with every headcount. The second is AI only: speed and scale, but no judgment when judgment is the entire point — and, increasingly, customers who don't want it. In one analysis of customer-service interactions, 71 percent of people said they would rather deal with a human, and 60 percent said chatbots routinely fail to understand their issue. In financial contexts the trust gap is even stronger: in the same research, just 9 percent of people were comfortable with a company using their financial information, against 47 percent for their purchase history. A separate 2025–26 study found that 84.7 percent of consumers would prefer a human, and that 80 percent still would even when promised the AI could resolve their issue.
Notice the asymmetry, because it is the whole strategy. For simple, fast, low-stakes tasks, customers increasingly prefer automation — they don't want to wait on hold to reset a password. For complex, sensitive, high-stakes moments — exactly the ones that decide loyalty and revenue — they want a capable human with the full context already in front of them.
The correct answer was never people or AI. It is AI where speed wins, people where trust decides, engineered together so each makes the other better. Not hybrid. Not layered. Engineered together.
Here is another story worth sitting with: 81 percent of consumers believe AI in customer service is deployed mainly to cut costs, not to improve their experience. They are, frequently, right. And that is the deeper failure — not the technology, but the intent behind it.
The commercial reckoning has already started
Here is where it gets uncomfortable for a lot of providers, and clarifying for buyers.
For two decades, the dominant commercial model in this industry has been the seat: you pay for people occupying your account, by the hour, regardless of what the work returns. AI is dismantling that logic in real time. Analysts now describe the move from seat-based to outcome-based commercial models as the most significant pricing transformation in the history of outsourcing, with Deloitte projecting seat-based revenue share to keep falling through the decade.
But two cheaper proxies have rushed in to fill the gap, and neither is accountability. The first is activity dressed up as value — billing for machine effort, where an AI agent that loops fifty times before failing still rings the register fifty times. The second is the structural conflict baked into legacy providers whose revenue depends on seats: the better their AI performs, the fewer seats you need, and the more it erodes their own model. As one set of industry observers put it, if a seat-based provider pitches you an AI agent, the fair question is how much your bill is going to shrink.
Meanwhile the accountability itself is rare. McKinsey's 2025 work on AI found that only 39 percent of organisations could attribute any EBIT impact to their AI at all, and that across software vendors, a tiny share have moved to genuine outcome-based commercials (McKinsey, via CustomerThink). Everyone is talking about outcomes. Almost no one is measuring for or being paid on them.
This is the difference between accountability and theatrics. Theatrics is announcing how much AI you shipped. Accountability is committing to what the experience does to revenue, margin and trust — and being measured against it.
What "owning the outcome" actually requires
If the problem is fragmentation, tunnel-vision metrics, missing intelligence and a commercial model that rewards effort over results, then the answer can't be another tool bolted onto the pile. It has to be a system — one that is designed, run and owned end to end, by one party willing to be held to the numeric business impacting results.
That is the shift we build for at Mplus, and the category we think it creates deserves a new name: Experience Intelligence. Not a label, and not "AI plus people" with a new coat of paint — a closed-loop operating system in which every interaction is captured as structured data, interpreted into usable intelligence, acted on by automation or human judgment as the moment demands, delivered across every channel, and fed back so the next interaction is better than the last. People-led where trust decides. AI-led where speed wins. Built on AI developed inside live operations, because that is the only place it learns to hold up under real volume and real risk.
Two structural advantages make that accountability real rather than rhetorical.
The first is ownership. One system, one accountable owner, performance tied to the P&L — instead of five vendors and five handoffs where the outcome belongs to no one.
The second is governance, and in a European context it is fast becoming a hard commercial filter rather than a footnote. The EU AI Act is now in force, with prohibitions live since February 2025, general-purpose AI obligations since August 2025, and high-risk obligations — the category that includes regulated uses such as credit scoring — phasing in next, carrying penalties that exceed those of GDPR. Crucially, the regulation reaches any organisation whose AI touches people in the EU, even those merely procuring it. For a regulated business, "where does the intelligence run, who governs it, and can you prove it" is becoming a question legal and procurement ask before anything else. An EU-headquartered, EU-governed system where your data never has to move is not a technical detail. It is a risk-management argument — and one that non-EU, template-driven models can't answer by policy alone.
The question to take into your next RFP or vendor meeting
Customer experience has always had a cost. The opportunity in front of every serious brand right now is to make it finally have a return — and to stop accepting partners who can only vary how many people sit on the account while you're trying to manage an outcome.
So before your next review, do one thing. Write down the three business outcomes you actually need — in revenue, in retention, in risk. Then make every provider answer to those, not to a price per seat. Ask them for a number they will stand behind.
The ones who change the subject back to efficiency are telling you exactly where their model ends. The ones who can answer are the ones building the future of this industry — where, when your experience is intelligent, it finally earns its place on the P&L.