Centene's AI Strategy: How the Medicaid Leader Is Rethinking Member Experience in 2026

TL;DR#

Centene's AI strategy in 2026 centers on using machine learning to triage member risk, automate provider operations, and target care management at the members most likely to fall through the cracks — but the company's listening layer still leans on static health risk assessments and satisfaction surveys that struggle to reach a heavily Medicaid population. Centene is the largest Medicaid managed-care organization in the U.S., with roughly 27.6 million total members at year-end 2025, including about 12.5 million Medicaid members, and full-year 2025 revenue of $194.8 billion (Centene 2025 results). Its predictive models, including the HALO risk program, identify who needs help; they cannot explain why a member misses appointments, ignores a renewal notice, or distrusts the system. For Medicaid members facing transportation gaps, language barriers, and unstable housing, the "why" is the whole game — and most disenrollments are procedural, not eligibility-driven (Commonwealth Fund). Conversational AI interviews close that gap by letting members describe their barriers in their own words, at scale, without a form standing in the way. This is the layer Centene's analytics-heavy stack is missing.

Who Centene serves and why member experience is hard#

Centene is the largest Medicaid managed-care organization in the United States, serving a member base that is among the hardest to reach in American healthcare. At year-end 2025 the company reported roughly 27.6 million total members, including about 12.5 million Medicaid members, just under 6 million Health Insurance Marketplace members, and approximately 8.7 million Medicare prescription-drug members, on full-year 2025 revenue of $194.8 billion, up from $163.1 billion in 2024 (Fierce Healthcare). It is the number-one carrier on the federal Marketplace and holds the largest concentration of Dual Eligible Special Needs Plan (D-SNP) members among its peers.

That scale comes with a structural challenge: government-program members are disproportionately affected by social determinants of health (SDOH) — unreliable transportation, food and housing insecurity, limited English proficiency, and low trust in institutions. The federal government defines these as the conditions in which people live, work, and age, and they shape health outcomes more than clinical care alone. For a Medicaid member, a missed appointment is rarely indifference; it is more often a transportation gap. Non-emergency medical transportation is a mandatory Medicaid benefit precisely because, without it, many beneficiaries simply cannot get to care.

This is the population Centene's member-experience strategy has to serve — and it is exactly the population that traditional surveys and intake forms fail. The same dynamic plays out across the payer industry, which is why the largest health insurer's member-experience strategy and Cigna's conversational care-navigation approach are converging on the same problem from different angles.

What is Centene's AI strategy in 2026?#

Centene's AI strategy in 2026 is to apply machine learning across three layers — provider operations, member risk stratification, and care-management targeting — to contain medical costs and improve outcomes for a predominantly government-sponsored member base. The strategy is analytics-led: AI tells Centene who is at risk and what to automate, while human care managers and outreach teams handle the relationship.

On the operations side, Centene has said it is deploying AI within provider-contracting operations to reduce the manual labor of configuring and maintaining contracts, enabling faster, more automated provider payments. The company frames AI as a lever to "drive quality and efficiency" across the business, with these initiatives accelerating through 2025 (Centene SEC filing).

On the clinical side, Centene leans on predictive analytics to flag rising-risk members. Its HALO program, an internal risk model, has reported roughly 98% accuracy in identifying high-risk conditions and pinpointing which interventions would help a given member, then routing them to the appropriate program (Emerj analysis of Centene AI). Care managers then meet the highest-risk members to assess needs and get ahead of barriers to care. This is genuinely strong infrastructure — and it mirrors the playbook other large carriers run. What it shares with them is a blind spot: the model can score a member's risk without ever capturing the member's own account of their situation.

The 2025 pressure test: why member understanding became a margin issue#

Centene's member-understanding gap stopped being an abstraction in 2025, when mispriced acuity turned into real financial pain. Centene reported an $11 billion GAAP loss in the fourth quarter of 2025, driven by elevated medical costs and Medicaid-rate misalignment following the historic eligibility-redetermination cycle, during which the company navigated membership and acuity shifts across 30 states (Fierce Healthcare). CEO Sarah London told investors the company was "laser-focused" on improving the Medicaid business and aligning rates with the post-redetermination book.

The redetermination episode is the clearest illustration of the listening gap. As pandemic-era continuous-enrollment protections ended, millions of members had to actively re-verify eligibility. The Commonwealth Fund found that most Medicaid disenrollments are procedural — members who are still eligible lose coverage because they never received the renewal notice, did not understand the process, or did not respond within a tight window (Commonwealth Fund, June 2025). Children's churn rose across racial and ethnic groups after annual renewal, with Hispanic children seeing the largest increase, reflecting greater barriers to staying covered.

A satisfaction survey cannot surface a barrier the member never reported. A risk model cannot explain why a renewal notice went unread. The "why" behind procedural churn lives in lived experience that static instruments don't capture — and that gap shows up on the income statement as mispriced acuity and avoidable disenrollment. This is the same structural failure we cover in why AI-first customer research cannot start with a web form.

Where forms and surveys bottleneck Medicaid member understanding#

Static health risk assessments (HRAs) and satisfaction surveys bottleneck Medicaid member understanding because they front-load effort onto exactly the members least able to absorb it. The traditional listening stack assumes a member who reads English at a high literacy level, has stable contact information, trusts the sender, and will translate a complex life into checkboxes. For a large share of Medicaid members, several of those assumptions fail at once.

The failure modes compound:

• Low completion among the highest-need members. Survey and HRA response is lowest exactly where understanding matters most — among members with unstable housing, language barriers, or low trust. The data that comes back is biased toward the members who were already easy to reach.
• No follow-up on the moments that matter. A member who checks "transportation is sometimes a problem" never gets asked the follow-up that would route them to non-emergency medical transportation. Forms capture a field; they cannot probe.
• Procedural barriers stay invisible. Research consistently shows that members who get live, in-person help are nearly twice as likely to complete enrollment, and that targeted outreach prevents procedural disenrollment (Icario on Medicaid engagement). A static form is the opposite of live help.
• SDOH gets flattened into a code. A "Z-code" for housing instability tells a care manager a category, not the member's actual constraint — whether they need a ramp, a shelter referral, or just a ride.

The same survey-versus-conversation tradeoff that erodes data quality in commercial research is amplified in Medicaid, where the gap between a checkbox and reality is widest. We unpack the mechanics in why conversations beat surveys for real customer research and in the broader case for an AI survey alternative built around conversation. For care and member-experience teams specifically, moving beyond NPS-style scores to the reasoning behind them is the difference between a number and an actionable insight.

How conversational AI interviews capture the "why" behind care gaps#

Conversational AI interviews capture the "why" behind care gaps by letting members describe their situation in their own words, at scale, while an AI interviewer follows up on vague or high-signal answers in real time. Instead of a member translating their life into a dropdown, the member talks; the AI listens, probes, and routes — the same way a skilled care manager would, but across hundreds of thousands of members at once.

Concretely, here is what that unlocks for a Medicaid-heavy book:

1. Barrier discovery before the care gap appears. An AI interviewer can ask an open question — "What's getting in the way of your next appointment?" — and follow up on the answer. A member who says "it depends on whether my cousin can drive me" reveals a transportation barrier a checkbox would have flattened. That maps directly to a non-emergency medical transportation referral.
2. Procedural-churn prevention in the member's voice. Ahead of a renewal window, a concierge-style intake agent can check whether the member received the notice, understood it, and knows their deadline — surfacing the misunderstandings that drive procedural disenrollment before coverage lapses.
3. Continuous listening instead of an annual pulse. Because an AI interviewer agent runs conversations on an ongoing cadence rather than once a year, acuity and barrier shifts surface as they happen — the exact signal mispriced in the 2025 redetermination cycle.
4. Language and literacy met where the member is. A conversational agent can adapt to plain language and follow up to confirm comprehension, rather than assuming the member parsed a dense form correctly the first time.

This is the model Perspective AI is built on: AI-powered customer interviews at scale that replace the form with a conversation. The same approach that lets product and CX teams capture intent is what lets a payer capture a member's actual barrier. The pattern is well established in adjacent verticals — see how conversational AI reshaped the insurance member relationship at Lemonade and how telehealth networks replaced patient-intake forms with conversation. For a Medicaid leader, the prize is bigger: better-understood acuity, fewer procedural disenrollments, and care management aimed at the real barrier rather than a proxy code.

The Medicaid context: why this matters more than in commercial healthcare#

Member understanding matters more in Medicaid than in commercial healthcare because the consequences of a missed signal are higher and the cost of reaching members is structurally greater. A commercial member who skips a survey is a data point lost; a Medicaid member whose barrier goes uncaptured may lose coverage, miss preventive care, and re-enter the system later, sicker and more expensive.

Three dynamics make the stakes asymmetric:

• Outcomes are barrier-driven, not preference-driven. In commercial care, satisfaction often turns on convenience. In Medicaid, outcomes turn on whether the member can physically get to care, afford the gap, and trust the messenger — SDOH factors that the federal government and researchers identify as primary drivers of health disparities (KFF on Medicaid and health equity).
• Churn is mostly avoidable and mostly procedural. Because most disenrollments are procedural rather than eligibility-based, better understanding directly translates into retained coverage and stabilized acuity — a lever that shows up in both outcomes and margin.
• The members who most need help are the hardest to survey. The bias in traditional listening is not random; it systematically misses the highest-acuity members, which is precisely how acuity gets mispriced.

For teams thinking about how this fits a modern stack, our guide to the customer-research tools modern product and CX teams actually use and the framework for AI-powered customer experience from first touch to renewal both apply directly to payer member journeys. Member experience and member research are a job built for CX teams, and the listening cadence is the same one we describe in how to build a voice-of-customer program from scratch.

What a conversational listening layer would add to Centene's stack#

A conversational listening layer would sit alongside Centene's existing predictive models, not replace them — feeding the human "why" into a stack that today is strong on the statistical "who" and "what." HALO and the company's predictive analytics already identify rising-risk members with high accuracy; a conversational layer answers the question those models can't: what is actually standing between this member and their care.

Other large carriers are wrestling with the identical gap, which is why surveying the field — from the largest health insurer's 2026 member-experience moves to Cigna's 190M-member care-navigation strategy — points to the same conclusion: the analytics arms race is maturing, and the remaining edge is in capturing the member's voice at scale. The market category forming around this is worth understanding; we map it in our roundup of the best AI customer-insight platforms for the enterprise in 2026.

Frequently Asked Questions#

What is Centene's AI strategy?#

Centene's AI strategy applies machine learning across provider operations, member risk stratification, and care-management targeting to contain costs and improve outcomes for its largely government-sponsored member base. It includes AI in provider-contracting operations and predictive risk models such as the HALO program, which has reported roughly 98% accuracy in flagging high-risk conditions. The strategy is analytics-led, using AI to decide who needs help and what to automate while human teams manage the relationship.

How many members does Centene have?#

Centene reported roughly 27.6 million total members at year-end 2025, making it the largest Medicaid managed-care organization in the United States. That total includes about 12.5 million Medicaid members, just under 6 million Health Insurance Marketplace members, and approximately 8.7 million Medicare prescription-drug members. Centene is also the number-one carrier on the federal Marketplace and holds the largest concentration of Dual Eligible Special Needs Plan members among its peers.

Why is member experience harder for a Medicaid plan?#

Member experience is harder for a Medicaid plan because the population is disproportionately affected by social determinants of health — unreliable transportation, housing and food insecurity, language barriers, and low institutional trust. These factors drive health outcomes more than clinical care alone and make members structurally hard to reach with surveys or forms. The members who most need help are systematically the hardest to survey, which biases traditional listening data toward the members who were already easy to reach.

How can conversational AI improve Centene member research?#

Conversational AI can improve Centene's member research by letting members describe their barriers in their own words at scale, while an AI interviewer follows up on high-signal or vague answers in real time. This surfaces the "why" behind care gaps and procedural disenrollment — the context that static health risk assessments and satisfaction surveys miss. It enables pre-visit barrier discovery, renewal comprehension checks, and continuous listening rather than an annual pulse, feeding actionable member context into Centene's existing predictive models.

What caused Centene's 2025 financial difficulties?#

Centene's 2025 financial difficulties stemmed primarily from elevated medical costs and Medicaid-rate misalignment following the historic eligibility-redetermination cycle, contributing to an $11 billion GAAP loss in the fourth quarter. As pandemic-era continuous-enrollment protections ended, membership and acuity shifted across 30 states faster than rates could adjust. CEO Sarah London described the company as "laser-focused" on improving the Medicaid business and aligning rates with the post-redetermination book.

Conclusion#

Centene's AI strategy has built a strong statistical engine — HALO and its predictive models reliably tell the largest Medicaid managed-care organization in the country who is at risk and what to automate across 27.6 million members. What that engine cannot do is explain why a member misses an appointment, ignores a renewal notice, or distrusts the plan. For a population shaped by transportation gaps, language barriers, and unstable housing, the "why" is where outcomes and retention are won or lost — and the 2025 redetermination cycle proved that the cost of missing it shows up directly in mispriced acuity and avoidable disenrollment.

Static health risk assessments and satisfaction surveys cannot capture that context, because they ask the hardest-to-reach members to translate complex lives into checkboxes. Centene's AI strategy would be meaningfully stronger with a conversational listening layer that captures the member's voice at scale and feeds it into the models it already runs. That is exactly what Perspective AI is built to do: run AI-powered member interviews at scale that replace the form with a conversation, follow up on what matters, and surface the barriers behind the data. To see how conversational AI interviews fit a payer's member-experience program, explore Perspective AI's interviewer and concierge agents and start capturing the "why" your current surveys are missing.