The 5 Stages of Payer CX Maturity, and Why Most Plans Get Stuck in the Middle
Originally Published by Simplify Healthcare
By Vishal Bhabad, Vice President of Growth, Xperience1™
When a health plan’s AI isn't working, the instinct is to blame the technology. Wrong model, wrong vendor, wrong use case. But in most cases, the AI itself isn't the problem. The problem is what the AI was built on top of, and the fact that nobody fixed that first.
To understand why, it helps to understand where your plan actually sits today, and what it truly takes to move forward. That's what the CX Maturity Scale was built to show.
The 5 Stages of Payer CX Maturity
We created this framework because we kept having the same conversation with payer leaders across the country. Everyone wanted to talk about AI and personalization and predictive member engagement. But when we got into the details of how their operations actually worked, it became clear that most plans were trying to solve a stage four or five problem without having the stage two or three foundation in place.
The CX Maturity Scale maps five stages of member experience capability, from reactive and call-center-first to fully predictive and AI-driven. The scale was derived from various public sources alongside our years of experience working closely with payers and understanding current state and challenges at the CX side.
The scale breaks down like this:
Stage 1: Reactive. This is the call-center-first model most health plans were built around, where data lives in silos and members have to work hard to get a simple answer, often waiting fifteen minutes, getting transferred, and starting over from scratch.
Stage 2: Standardized. The plan has launched a portal and mobile app and service is more consistent, but the underlying data is still scattered across disconnected systems, so the digital experience looks better without actually being able to answer the questions members are asking.
Stage 3: Integrated. Claims, clinical, and benefits data are unified into a single coherent foundation, and a member can move from a portal to a phone call without repeating themselves, because the systems finally share a common picture of who they are and what's happened in their care.
Stage 4: Personalized. With a clean integrated foundation underneath, the plan can start using AI-driven nudges and proactive outreach to reach members before they have to reach out themselves, mapping journeys to specific member segments rather than waiting for problems to surface.
Stage 5: Predictive. This is real-time decision intelligence, where the plan anticipates member needs before any contact happens at all, and it's the destination most payer leaders are describing when they talk about what they ultimately want AI to do for their organization.
Where Most Plans Actually Are
Most health plans today sit somewhere between stages two and three. They have a portal. They have an app. They may even have a chatbot. But the data feeding those tools is still fragmented across systems that were never designed to talk to each other.
Think about what actually happens when a member asks a simple question: why did my claim fail?
That question touches claims data, benefits data, clinical criteria, and provider information. Each of those lives in a different system, formatted differently, governed differently, with different logic for how it gets interpreted. A Facets table doesn't automatically translate into a plain English explanation, and a benefits document sitting in a data warehouse has no way of knowing it's related to the claim that just processed. That means someone on the service team has to manually log into multiple systems, piece together an answer from whatever they can find, and hope that what they've assembled is accurate enough to actually help the member.
That fragmentation is why call handle times run fifteen minutes or more, why portals still can't self-serve the most common questions members are asking, and why members end up calling back after they've already tried to find the answer on their own. None of this is a people problem or a process problem. It is a data foundation problem, and no amount of training or tooling fixes it without addressing what's underneath.
The Danger of Skipping a Stage
Here's where it gets important, and where a lot of plans are quietly in trouble right now.
Over the last few years, boards pushed hard for AI adoption, regulations created real urgency around digital member experience, and vendors showed up with demos that made personalized and predictive member engagement look achievable in a matter of months. Given all of that pressure, a significant number of health plans did something completely understandable and deployed AI on top of systems that were never designed to support it.
The result is predictable in hindsight, because you cannot put intelligent automation on top of fragmented, unvalidated data and expect it to produce accurate and trustworthy answers. What you get instead is the wrong answers delivered faster, chatbots that confidently tell members incorrect things, automated workflows that route claims to the wrong place, and member experiences that somehow feel worse than just calling in, because at least on the phone someone eventually pieces together an answer.
We are seeing this play out in real conversations with clients. Plans that jumped from stage two to stage four are coming back saying their chatbot isn't doing what it was supposed to do. Their automated responses aren't accurate, but they don't know how to fix it because the agents are already deployed and members are already using them. Undoing it is harder than building it right the first time would have been.
The maturity scale exists to make this visible before it happens. You cannot go from reactive to predictive overnight. You can try, but skipping stages doesn't save time. It creates a different, more expensive problem.
What Has to Be True at Each Stage
The reason plans get stuck between stages two and three comes down to one thing: their systems don't share a common language.
Claims information sits in one tool, benefits live in another, and clinical data is somewhere else entirely. None of these were designed to interpret each other's outputs or create the kind of shared context that makes a coherent member experience possible. What's missing is the layer that takes a claim status, connects it to the benefit rule that drove the outcome and the clinical policy that applied, and surfaces all of that in plain English to whoever is trying to answer the question. Without that foundation, you can add tools and channels and AI, but the experience stays broken no matter what you layer on top.
What Integration Actually Makes Possible
When plans get to stage three, something real changes that goes beyond the data simply being cleaner or more organized. What actually shifts is the plan's ability to make decisions across information that previously lived in complete isolation from one another.
A service rep can answer a question about a denied claim, the benefit rule behind it, and the appeal process all in one conversation without putting the member on hold to log into three different systems. A portal can give a member a meaningful answer to why something wasn't covered instead of pointing them to a phone number. And an AI agent can be trusted to reason across all of that without producing answers that constantly require human correction.
That's the foundation that makes stages four and five real. AI-driven nudges and proactive outreach require knowing enough about a member to reach out with something relevant. Predictive intelligence requires a data foundation accurate enough to build predictions on. Neither of those is possible when the underlying data is fragmented and unvalidated.
The plans that get to personalization and prediction reliably are the ones that did the work at stage three first.
What to Take Away From This
If you lead operations at a health plan, the most useful thing this framework can do is give you an honest picture of where you are and what it actually takes to move.
Most plans are sitting at stage two today, and while some have made it to stage three, very few are operating at stage four in any meaningful way. The ones who believe they are but skipped the integration work at stage three are finding out the hard way that the gap matters enormously.
The question to ask isn't whether your plan has AI, but whether your plan has the data foundation that makes AI trustworthy and whether your systems share enough common context that an AI agent can reason across them accurately. Put more simply: did you build stage three before you tried to build stage four?
If the answer is no, the right move isn't to slow down. It's to stop adding tools on top of a fragmented foundation and start fixing the foundation. Because the plans that get there first, that build the integrated data layer that makes intelligent experience possible, are the ones that will have the advantage when AI actually delivers on what it's been promised to do.
The CX Maturity Scale doesn't exist to make anyone feel behind. It exists to make the path forward legible. And the first step on that path is knowing honestly where you are.
Comments