TL;DR

• Gartner, Forrester, and Qualtrics XM Institute each publish CX maturity frameworks. Different labels, different scope, same finding: most programs cluster at the bottom. Seventy-one percent sit in the first two stages. Very few reach the top.
• The usual explanations (bad tooling, no exec buy-in, siloed teams) are real. They're also symptoms. The deeper problem is structural: programs built to measure, not to signal.
• Level 2 to Level 3 is an ownership problem. Level 3 to Level 4 is an intelligence problem. That second jump is the one nobody explains well.
• Signal-classified feedback (what's changing, what's at risk, what customers are asking) is the mechanism that connects standardized measurement to cross-functional action.
• Level 5 isn't a software purchase. It's what Level 4 looks like two years later, if the foundation holds.

If you collect customer feedback, you probably have a reasonable read on your customer experience maturity levels. Forrester has an opinion. Gartner has a model. Your quarterly NPS deck has a number that everyone nods at and nobody acts on.

What none of them tell you is what actually changes between levels. Not the labels. Not the descriptions. The mechanism. The thing that moves a program from "we collect feedback" to "feedback changes how we operate."

That's the gap. Not awareness. Not ambition. The gap is mechanical: most programs spend years at Level 2, collecting data across channels, analyzing it manually when they analyze it at all, and producing reports that describe what happened without telling anyone what to do about it.

The diagnosis has never been the hard part. There are five well-documented frameworks, a dozen maturity assessments, and more benchmarking surveys than any team has time to fill out. The problem is the prescription. "Get executive buy-in" and "break down silos" are true. They're also not specific enough to be useful when you're the one trying to move the program forward on a Tuesday afternoon.

This piece is about what moves. What changes at each transition, what blocks it, and why the hardest jump has less to do with organizational politics and more to do with what your feedback infrastructure was built to surface.

Every Major Framework Agrees, and Stops at the Same Place

Gartner, Forrester, and Qualtrics XM Institute have each published CX maturity frameworks. They measure slightly different things. Gartner focuses on CXM capability broadly. Forrester maps CX strategy and competitive differentiation. Qualtrics XM Institute tracks organizational culture maturity. The labels are different. Forrester uses four levels; the other two use five. None of them map neatly onto each other.

But strip the labels away and the shape is identical. Every framework describes the same arc: reactive and fragmented at the bottom, structured and owned in the middle, predictive and embedded at the top. And every framework finds the same uncomfortable distribution.

Gartner's research found 41% of organizations sitting at the most basic, fragmented stage, with sharp drop-offs at every level above. Qualtrics XM Institute's 2024 State of CX Management research put 71% of practitioners in the bottom two stages of maturity, with 41% still at Stage 1. Just 2% reached the highest stage. Forrester's 2026 CX predictions went further: budget pressure will lure 15% of CX teams into what they call a "death spiral," doubling down on metrics that no longer justify the program's existence.

Three frameworks. Three different lenses. One conclusion: most programs are stuck near the bottom, and the distance between where teams sit and where they need to be is not closing on its own.

Here's what none of them explain clearly. They describe each level well. What they don't describe is what structurally changes between levels. And specifically, why the jump from Level 2 to Level 4 is so much harder than it looks on a maturity curve.

The Real Reason Programs Stall Isn't What You Think

The conventional explanation goes like this: leadership doesn't prioritize CX, teams are siloed, the tooling is fragmented. All true. None of it is the root cause.

The root cause is quieter. Most CX programs are built to measure. Almost none are built to signal.

That distinction matters more than it sounds. A program built to measure tells you what the score is. A program built to signal tells you what's changing, what's at risk, and what's driving the change. One produces a dashboard. The other produces action. Most programs produce the dashboard and hope the action follows. It rarely does.

Forrester named this precisely in their 2026 predictions. CX teams, they wrote, are trapped in a stable but dysfunctional orbit, circling what amounts to measurement without meaning. The dashboards are full. The narrative is empty. The reports don't reveal which problems to solve, how to address them, or why they matter to the business.

Here's what that looks like on a regular Wednesday. A Level 2 program runs a quarterly cycle that goes something like this:

• Someone pulls the NPS data and compiles a report.
• Someone presents it to leadership.
• Everyone agrees the score should go up.
• Three months later, the process repeats. Nothing changed.

The report answered "what is the score" but not "what is changing within the score, what's driving it, and who specifically should do something about it."

That's not a governance failure. It's an infrastructure failure. The program was never designed to surface signals. It was designed to produce measurements. And measurement, however accurate, doesn't route itself to the person who can act on it.

Forrester's research found that more than 60% of CX professionals say their organizations lack formal closed-loop processes for feedback. Not because they don't care. Because the infrastructure that would make it possible, the layer that classifies feedback into signal types and routes it to the right people, was never built. Customers notice. Gartner research cited by CX Dive found only 16% of customers strongly believe their feedback drives actual change. That skepticism isn't irrational. For most voice of customer programs, the customers are right.

That layer, the one that classifies and routes, is what separates Level 2 from Level 4. And it's the piece most maturity frameworks describe as an outcome without explaining how to build it.

What the Jump at Each Level Requires

Each transition on the maturity curve has one dominant shift. One thing that, if you get it right, pulls the rest forward. Understanding which shift belongs to your level is what turns "we need to mature" from a slide title into something you can execute.

Level 1 to Level 2 is the easiest jump, and the only one where tooling really is the blocker. Pick a metric. NPS, CSAT, CES, whichever fits your program. Pick a channel. Send consistently. The barrier at Level 1 is inertia, not complexity. Most teams stuck here know it and already have a plan.

Level 2 to Level 3 is where most programs stall. And the blocker is almost never the platform. It's ownership.

A Level 3 program needs one accountable person with the mandate to standardize: agree on a shared question taxonomy, set a cadence, and build a real follow-up process for at least the worst responses. These are organizational decisions. No platform makes them for you.

This is why teams that go platform-shopping to solve a Level 2 problem tend to stay at Level 2. They've bought better measurement infrastructure for a program that still has no owner and no agreed standard. The data gets cleaner. The plateau remains.

The enterprise survey maturity model covers this transition in detail: the five dimensions where most programs have uneven maturity, and why your weakest dimension sets your ceiling.

Level 3 to Level 4 is the jump nobody explains well. And it's the one where signals become the lever.

A Level 3 program has standardized data. Surveys go out consistently. Responses are comparable across time and teams. Someone owns the program. That's genuinely valuable. But the data it produces is still a pile of responses that someone has to manually read, interpret, and route to the right person.

A Level 4 program has classified data. Feedback has been automatically sorted into operational signal types:

• What's driving experience this period
• What's newly emerging
• What's flagged as a retention risk
• What customers are asking rather than complaining about
• Where verbal sentiment contradicts the numerical score

Those classifications are what allow feedback to route to operations without a human analyst deciding it should.

Without that classification layer, Level 4 work (routing feedback to account managers, alerting product to an emerging theme, triggering a retention play when churn risk spikes) requires someone to sit between the data and the people who need it. Most teams fill that gap with spreadsheets. The ones that actually reach Level 4 don't. They've built or adopted a signal layer that does the classification automatically.

This is where the "advanced analytics" that big enterprise programs invest in becomes legible. What Gartner, Forrester, and the large platform vendors are all pointing at, in different language, is the same structural upgrade: from a program that stores feedback to one that classifies and routes it. If you're building the case internally, read more on enterprise survey ROI.

Level 4 to Level 5 depends entirely on having clean, integrated, historical data from Level 4. Prediction sits on top of classification and integration. You can't shortcut to it. The teams that reach Level 5 got there by doing boring Level 3 work first and holding the standard long enough for the data to compound.

What Changes When You Add a Signal Layer, at Every Level

Everything above has been conceptual. Signals are the lever. Classification is the mechanism. But what does that look like for a team sitting at Level 2 today versus a team already running a standardized program at Level 3?

The answer depends on where you start. The same infrastructure upgrade produces a different outcome at each level. Here's what changes.

If you're at Level 1 (ad-hoc, reactive): Signals won't help you yet. Your problem is more basic: you don't have consistent data to classify in the first place. The first move is establishing a cadence and a channel. Get to Level 2 first. That's a one-month project, not a six-month initiative.

If you're at Level 2 (siloed, manual analysis): This is where the shift hits hardest. Today, feedback comes in across channels, sits in separate tools, and gets read manually. If it gets read at all. Adding a signal layer means the feedback that used to live in spreadsheets now gets automatically classified:

• What's a churn risk
• What's an emerging complaint
• What's a product suggestion
• What's a question your help docs should be answering

You skip the weeks of manual tagging and go straight to knowing what changed and what to do about it.

Here's the thing about Level 2 teams that add signal classification: many of them jump to Level 3 and Level 4 in the same move. The classification solves both problems at once. Shared taxonomy? It happens automatically when the AI classifies by theme. Cross-functional routing? That's what happens when signals go to Slack, email, tasks, and tickets without a human deciding they should.

If you're at Level 3 (standardized, one owner): Your data is clean and comparable. You've done the hard governance work. What's missing is the layer between "we have the data" and "the right person is acting on it." Signal classification turns your standardized measurement program into an operational one:

• Churn signals surface in the same week they appear, not in next quarter's report.
• Product suggestions get routed to the product team the same day.
• The CS lead doesn't wait for the monthly review to learn about a complaint spike; they see it in a Slack digest that morning.

The gap between having data and using data? That's what closes.

If you're at Level 4 (integrated, cross-functional action): You're already acting on feedback. The signal layer sharpens what you act on. Mixed signals, where a customer gives a high NPS score but writes a negative comment, get flagged automatically instead of hiding in the aggregate. Emerging themes that would have taken another quarter to surface show up within the current reporting period. That kind of sentiment analysis on customer feedback is what moves a program from responsive to anticipatory.

Here's the thing: what large enterprise platforms have charged six figures a year for (automated theme detection, churn risk flagging, intent classification, temporal comparison) is the same capability that mid-market programs now have access to through AI customer feedback analysis. The five signal types that matter most:

• Key drivers: what's influencing experience right now
• Churn signals: where retention risk is concentrated
• Feedback queries: what customers are asking and suggesting
• Praise and recommendations: what's working and who's advocating
• Mixed signals: where scores and sentiment contradict each other

Those aren't a feature list. They're the classification layer that makes closing the feedback loop operationally possible rather than aspirationally nice.

And that matters for the maturity conversation because most frameworks describe Level 4 as "integrated" without explaining what integration actually requires. It's not a data pipe between your survey tool and your CRM. It's classified, routed feedback arriving at the right person, in the right channel, with enough context to act. No one waiting for someone else to pull a report.

What This Means for Your Program

The frameworks are right that most programs are stuck. They're also right that the path up runs through governance and culture, not technology alone.

But here's where the honest part comes in. Governance and culture don't close feedback loops on their own. At some point, moving from Level 3 to Level 4 requires building the layer that turns a standardized measurement program into one that classifies feedback into signals and routes those signals to the people who own the metrics they affect.

That layer is no longer reserved for organizations with six-figure enterprise contracts and six-month implementations. It's the infrastructure requirement for the CX programs that will still be funded, and still be relevant, when Forrester's predicted budget pressure arrives.

The teams that will break free from the death spiral share one architectural difference from the ones that won't. They stopped treating feedback as something to store. They started treating it as something to classify, route, and act on. That shift doesn't happen by accident. It requires a deliberate decision to build the signal layer, and the right system to run it.