AUGUST 2026 | ISSUE 002
What Q2 Earnings Tell Us About the Health Plan the Future Requires
Synthesizing the signals shaping the next generation of health plan leadership. An executive briefing on the emerging operating model of health plans.
By John G. Murtha
Letter from the Editor
Seven publicly traded health plans reported second-quarter results by the end of July.
Most executives experienced those results as seven separate earnings calls: different
management teams, different portfolios, different analyst questions and different explanations
for the quarter.
I have treated them as one conversation.
Read together, the calls reveal more than the direction of medical costs or the quality of
quarterly execution. They show an industry attempting to redesign itself while continuing to
operate at full scale.
The pressure is familiar. Medical costs remain elevated. Government programs are less forgiving. Administrative costs remain stubborn. Technology environments remain fragmented. Providers and members expect decisions to move faster. Regulators increasingly expect the same.
The response, however, is changing.
Clinical operations are becoming a financial discipline. Artificial intelligence is entering production workflows. Prior authorization is being redesigned around automation, exceptions and provider performance. Clinical data is moving closer to the point of decision. Plans are becoming more selective about markets, membership and capital. Capabilities once developed inside departments are being reconstructed as enterprise infrastructure.
This edition synthesizes the recurring signals, identifies the patterns beneath them and considers their implications for executive leadership.
It does not ask which plan won the quarter.
It asks what the quarter taught us about the institution the health plan is becoming.
The answer is not a finished model. It is a direction of travel.
The future health plan will be less a collection of insurance functions and more a coordinated
decision enterprise: continuously sensing change, translating evidence into action and aligning decisions made across different operating horizons.
That transition is already under way.
Section I — Executive Intelligence - What Q2 Earnings Tell Us
Observation 1: Margin pressure has become structural.
Evidence
• Medical-cost ratios remained near or above 90% in several government-program lines of
business, including among plans that exceeded earnings expectations.
• Several management teams described recovery paths extending through 2028 or 2029
rather than expecting one annual pricing cycle to restore historical margins.
• Across government programs, management teams increasingly prioritized sustainable
unit economics over membership retention, exiting or retrenching where rates, acuity
and operating requirements could not be reconciled.
Implication
Margin recovery can no longer be treated principally as a pricing exercise. It requires
coordinated action across product design, clinical operations, network strategy, risk
adjustment, administration and market participation.
Observation 2: Clinical operations have become financial strategy.
Evidence
• One large-scale complex-care model reported inpatient admissions approximately 12%
lower and emergency-department use approximately 13% lower, with expected medical
savings in the mid-teens.
• Utilization management, transitions of care, site-of-service management and earlier
identification of clinical needs were repeatedly presented as margin-recovery levers.
• Leading plans are beginning to detect emerging medical-cost trends in days and hours
rather than waiting for completed claims and monthly reporting cycles.
Implication
Clinical operating performance is now an earnings variable. The agendas of the chief
medical officer, chief operating officer and chief financial officer are increasingly inseparable.
Observation 3: AI has entered execution.
Evidence
• Several multi-year AI programs now carry productivity ambitions measured in 100–200
basis points.
• Current deployments are concentrated in prior authorization, claims, payment integrity,
clinical-document review, care-gap identification, contact centers and member
navigation.
• Management teams are increasingly distinguishing between automating existing work
and redesigning the process itself.
Implication
The leadership question is no longer whether to experiment with AI. It is which workflows
should change, what work should disappear, where human authority remains and how the
clinical and financial effects will be measured.
Observation 4: Government programs continue reshaping enterprise priorities.
Evidence
• Medicare Advantage performance remains tied to pricing, Stars, risk-model changes,
provider performance and earlier condition identification.
• Medicaid redeterminations produced smaller, often higher-acuity populations while
intensifying pressure for adequate state rates.
• ACA exchange performance diverged materially, prompting some plans to expand,
others to retrench and others to exit.
Implication
Government programs are not operating at the edge of the enterprise. Their rules
increasingly determine data priorities, clinical workflows, market portfolios, technology
roadmaps and capital requirements.
Observation 5: Administrative simplification remains elusive.
Evidence
• Several plans improved administrative-cost ratios while establishing further multi-year
productivity targets.
• In advanced connected workflows, plans are reporting material reductions in avoidable
denials, manual contacts and authorization volume—not simply faster processing of the
same work.
• Yet manual documentation requests, delayed claims, fragmented clinical data, repeated
reconciliation and regional variation remain widespread.
Implication
Automation is producing local gains faster than the enterprise is becoming simpler. The next
stage requires retiring work, consolidating platforms and removing the organizational causes
of duplication.
Observation 6: Capital allocation is becoming more disciplined.
Evidence
• Several plans accepted near-term earnings pressure to fund provider connectivity, clinical
capabilities, platform modernization and operating-model change.
• Market exits and portfolio pruning were increasingly justified by the absence of
sustainable economics rather than membership considerations alone.
• Acquisitions were more often framed around missing capabilities—specialty care, home-
based services, clinical data and provider enablement—rather than revenue growth alone.
Implication
Capital allocation is becoming an operating-model decision. Boards must compare
repurchases, modernization, capability acquisitions and market exits within one enterprise
architecture.
Observation 7: Enterprise capabilities are replacing departmental organization.
Evidence
• Plans are centralizing selected clinical and administrative functions while preserving
necessary local market adaptation.
• Common data platforms, provider-connectivity networks, shared clinical policies and
enterprise AI environments are becoming recurring investment priorities.
• Management teams increasingly described data, automation and decision logic as
reusable enterprise assets rather than departmental tools.
Implication
The future health plan will not be built by making every department better in isolation. It will
be built by creating capabilities that allow the enterprise to sense, decide and act across
departmental boundaries.
AI The throughline
The quarter did not reveal seven unrelated strategies.
It revealed variations of the same structural response:
The health plan is moving from a portfolio of functions toward an enterprise
system of trusted decisions.
The implications extend beyond quarterly performance. Taken together, the signals point
to the operating model the future will require: one capable of detecting change earlier,
establishing shared evidence, producing trusted decisions and coordinating action across
functions that operate on different horizons.
That is the architecture now beginning to emerge.
The Emerging Operating Model
The health plan the future requires will depend on five connected capabilities.
1. SENSING
Recognize meaningful changes in utilization, acuity, provider behavior, clinical events,
policy and market economics while there is still time to act.
2. SHARED EVIDENCE
Normalize, reconcile and make traceable the clinical, claims, pharmacy, authorization,
quality, risk and provider information needed to support action.
Different functions may ask different questions. They should not begin with different facts.
3. TRUSTED DECISIONS
Apply consistent clinical and business logic through explicit decision rights, transparent
evidence and appropriate governance.
A trusted decision is explainable, context-aware, defensible and understood by the people
expected to act on it.
4. COORDINATED EXECUTION
Embed decisions where work occurs: in provider workflows, care teams, claims operations,
contact centers and member channels.
Automate the routine. Route exceptions to qualified people with the relevant evidence
attached.
5. CONTINUOUS LEARNING
Measure whether decisions improved cost, quality, experience or operating capacity.
Return outcomes to the system so that local learning becomes enterprise knowledge
The Structural Shift
FROM: A portfolio of functions TO: A coordinated decision enterprise
• Separate data • Shared evidence
• Departmental logic • Trusted decisions
• Sequential hand-offs • Embedded execution
• Local optimization • Enterprise capabilities
• Retrospective learning • Continuous learning
Transformation moves at the speed of trust
Organizations move slowly when leaders dispute the evidence, question the logic or lack
confidence in how a decision was reached. Shared evidence reduces debate about the
facts. Trusted decisions reduce hesitation, escalation and rework. Coordinated execution
allows the enterprise to move.Observed outcomes either reinforce that trust or reveal where
the system must change.
The health plan the future requires is less a collection of functions and more a system for
producing trusted decisions at enterprise scale.
SENSE → ESTABLISH EVIDENCE → DECIDE → EXECUTE → LEARN
Section II — Operator’s Observation — The Enterprise Doesn’t Operate on One Calendar
Health plans often speak as if the enterprise shares a planning horizon.
It does not.
Actuarial teams are pricing future populations. Product leaders are designing benefits for a year that has not begun. Sales teams are preparing for enrollment. Clinical operations are managing needs that are changing now. Risk adjustment and quality teams are validating care already delivered. Finance is closing a quarter. Technology is building capabilities that may not reach scale for several years.
Each function is rational within its own calendar.
The difficulty begins when the enterprise mistakes those calendars for one another.
A rate decision may have a twelve-to-twenty-four-month horizon. A clinical intervention may have a useful life measured in hours. A quality measure can affect care today, a Star Rating next year and revenue after that. A new data platform may require years of investment before it changes the cost structure.
These are not different descriptions of the same clock. They are different decision windows.
The answer is not to force every function onto one calendar. That would be impossible and undesirable.
The answer is to design the enterprise around the hand-offs between clocks.
A change in utilization detected through clinical operations must reach actuarial leadership before the next pricing decision closes. A new benefit design must reach member and provider workflows before the product takes effect. A diagnosis documented during an encounter must move through quality, risk and revenue processes without being reconstructed months later. A technology investment must carry an explicit path from platform delivery to workflow adoption and financial effect.
Most operating failures occur at these boundaries.
The plan may possess accurate information but receive it after the opportunity to intervene has passed. It may identify a care gap but fail to deliver it inside the clinical encounter. It may centralize utilization management without standardizing the evidence available to reviewers. It may automate one step of a process while leaving the surrounding work intact. It may approve a transformation program without assigning authority over the cross-functional changes required to realize its value.
The essential leadership task is synchronization. Not coordination in the ceremonial sense of meetings, committees and status reports.
Synchronization means that the output of one operating clock becomes usable input for another before its decision window closes.
That requires three disciplines.
First, define the decisions that matter.
“Improve interoperability” is not a decision. “Determine whether this request can
be approved without manual review” is.
“Modernize risk adjustment” is not a decision. “Determine which clinically
supported condition has not reached the encounter record” is.
Second, establish the maximum tolerable latency for each decision.
A quarterly report, a thirty-day claim, a same-day clinical event and a sub-minute
authorization are not interchangeable information products.
Third, assign accountability across the boundary.
When a clinical signal fails to influence pricing, the failure belongs neither solely
to clinical operations nor to actuarial. It belongs to the operating system
connecting them.
The enterprise does not need one calendar. It needs shared evidence capable of supporting trusted decisions across many calendars.
The clocks do not need to run at the same speed. They need to move the enterprise in the same direction.
Section III — Framework
Section IV — Working Paper Preview
Working Paper 003: The Decision Enterprise
The clocks illustrate the problem.
The Decision Enterprise will examine the operating system required to connect
them.
SIGNAL → EVIDENCE → DECISION → ACTION → OUTCOME → LEARNING
Three key insights
1. Data creates value only when it changes a decision.
The objective is not to accumulate information. It is to shorten the distance
between evidence and action.
2. Decision rights matter as much as analytical capability.
An enterprise may identify the correct action and still fail because authority,
accountability or workflow is unclear.
3. The enterprise learns only when outcomes return to the system.
Without a closed loop, every department and market repeats the same experiment.
Coming this fall.
Questions Leadership Teams Should Be Asking
1. Which portion of current margin pressure reflects inadequate pricing, and which
reflects delayed or inconsistent operating decisions?
2. Where do clinical, actuarial and financial teams rely on different versions of the
same member, provider or market reality?
3. Which AI deployments have removed work rather than accelerating one step inside
an unchanged process?
4. Where does information lose its value before crossing from one operating clock to
another?
5. Who owns decisions that cross clinical, actuarial, technology and market
boundaries—and do they possess the authority to change the system connecting
them?