The Future of Patient Experience

From dispersed feedback to structured system intelligence

The NHS listens more than ever before. Patient feedback is collected at unprecedented scale through surveys, digital channels, complaints systems, patient-reported measures, staff intelligence, regulatory submissions, and emerging AI tools. Yet volume alone has not produced the system-level learning the NHS needs.

The central question is no longer whether the NHS listens. It is whether it can reliably convert lived experience into clear, stable evidence that supports accountable improvement.

Recent maternity reviews, including the Ockenden Review of Nottingham University Hospitals NHS Trust and the National Maternity and Neonatal Investigation chaired by Baroness Valerie Amos, show why this distinction now matters nationally. The issue is not simply whether families, patients and staff have opportunities to speak. Voice is present, concerns are raised and patterns are visible. The failure occurs when those voices are not preserved as evidence, when evidence is not escalated, and when escalation does not result in accountable change.

The Amos investigation further sharpens this. Listening, hearing, and acting on concerns are not peripheral experience activities; they are critical safety functions. Listening data must be captured as safety intelligence, reviewed through governance, escalated to boards where patterns emerge, linked to measurable improvement and considered by regulators as part of safety assessment.

This summary article created by the Research and Development team at Akumen is informed by Akumen's sustained operational practice across more than sixty healthcare organisations, including over fifty NHS Trusts, sets out why the gap exists, what has sustained it, and what we believe it will take to close it.

The contradiction the system now must solve.

Patient narrative is trusted retrospectively to explain harm after inquiries begin, but it is not yet systematically trusted to prevent harm while there is still time to act. The future of patient experience depends on resolving that contradiction, not managing around it.

Five conditions that define the present moment

The full Akumen paper identifies five visible conditions in the current NHS patient experience landscape. None are failures of intent, they are structural consequences of operating under prolonged pressure.

1. Listening has expanded — but evidence has not

Feedback mechanisms have expanded, but expansion has not been matched by the methodological and governance infrastructure needed to treat narrative as evidence. Patient voice is collected, reported, and discussed — often sincerely — while structural change drifts. The system accumulates voices faster than it accumulates learning.

2. Fragmentation is visible but normalised

Patient experience data sits across multiple teams, systems, and reporting lines with no single mandate to integrate it. The hidden cost is signal dilution: patterns that would be visible at scale remain scattered and appear insignificant when confined to local ownership boundaries. In healthcare, what appears least often can sometimes matter most.

3. AI is being adopted on plausibility

Standard AI systems operate through probability-led inference: what appears most often shapes what is surfaced. In sparse, high-stakes narrative environments — patient safety records, complaints pathways, mental health assessments, coroner findings — this introduces a structural blind spot. Fluent outputs can be mistaken for verified evidence unless they are governed, validated, and constrained by human judgement.

4. Experience data sits upstream of litigation

The progression from concern to complaint to claim often begins with a patient or family sensing that something is wrong and that the system is not listening. Experience data can contain the earliest indications that harm is emerging long before a formal claim is initiated. The capacity to escalate has also changed: generative AI now helps patients and families produce more structured, legally framed correspondence.

5. Weak signals are present but not protected as system intelligence

Families describe them not being heard and staff raise concerns. Complaints, incidents, bereavement accounts, PALS contacts, debriefs, staff surveys and regulatory submissions all carry fragments of the same underlying picture. When these fragments remain separated, selectively summarised or trapped within local governance routes, the system may hear many things but still fail to learn.

The NHS does not lack patient feedback. It lacks an integrated meaning architecture capable of turning distributed voice into concentrated evidence.

Why has the system taken this shape?

The full Akumen paper examines five survival patterns that have accumulated under sustained pressure: structural survival mode, defensive routines, the relational entanglement of staff and patient experience, cultural fragmentation mirrored in data fragmentation, and the distortions introduced by self-selecting digital inspection models.

These are not failures of character. They are intelligent adaptations to a system under strain. A system that cannot combine its experience intelligence cannot learn quickly enough to prevent escalation, cannot prioritise with confidence, and cannot demonstrate improvement in a way that is robust and contestable. Acceptance of this is the precondition for redesign.

A particularly consequential pattern is institutional moral overload. Staff are asked to listen deeply to patients while lacking authority or resources to address upstream constraints. Clinicians hear distressing stories and managers see recurring themes. However if systemic barriers remain immovable, listening becomes emotionally costly rather than empowering. Asking for more feedback without giving people the power to act on it only adds to the strain.

The framework

The Akumen paper is structured around a four-stage model that mirrors the stages through which large institutions recover from sustained pressure.

AWARENESS	ACCEPTANCE	RECONSTRUCTION	RECONNECTION
See what is happening	Understand why it took this shape	Build what is missing	Assess whether it alters behaviour

Five architectural foundations

Reconstruction requires five specific design commitments; these are not aspirational principles they are design requirements. Organisations that have built them, even partially, demonstrate stronger signal detection and more confident board-level engagement with experience evidence.

Foundation	Design requirement
Narrative → Evidence	Structured, auditable, longitudinal analysis, with absence signals treated as first-class data.
Triangulation	Patient, family, staff, complaints, incidents, claims, safeguarding, operational, regulatory and board assurance evidence aligned as routine.
Governed AI	Semantic constraints defined before inference runs; human judgement protected as a deliberate firebreak.
Federated Meaning	Meaning integrated across organisational boundaries, not merely data access or platform interoperability.
Risk Infrastructure	Experience functioning as an early warning system upstream of harm, litigation, and reputational failure.

The question a frequency-optimised system asks is: what appears most often? The question patient safety demands are: what have we defined as mattering — and where is it present, however rarely?

What needs to change?

The direction of travel is clear. The gap between current state and required direction is not primarily a technology gap. It is a governance, methodology and culture gap.

Current state	→	Required direction
Listening as activity	→	Listening as evidence-generation
Fragmented silos	→	Integrated meaning architecture
AI adopted on plausibility	→	AI governed by semantic constraints
Frequency as proxy for importance	→	Consequence as the governing principle
Experience as commentary	→	Experience as risk infrastructure
Retrospective inquiry learning	→	Prospective narrative safety intelligence
Survival identity	→	Learning identity

The role of AI — getting it right

Artificial intelligence will expand its influence across summarisation, triage, trend detection, inspection modelling, and strategic forecasting. At scale, the architectural problem becomes systemic: if fluent outputs shape strategic framing without rigorous validation, confidence inflates faster than evidence, and attention is steered by what appears most often rather than what matters most.

The appropriate response is not to reject AI but to govern it. This means defining semantic constraints before inference runs — determining in advance what the system is required to search for, not just what it finds most often. It means testing outputs against rare but critical signals. It means maintaining human oversight as a deliberate firebreak, not a ceremonial step.

The National Maternity and Neonatal Investigation shows that AI can support human analysts in reviewing a large, time-limited body of evidence.

But this is very different from building stable, repeatable intelligence that boards and ICBs can track over time. One-off thematic analysis can help explain what happened, but it is one off . What is now needed is governed, auditable analysis that shows whether risk, culture, inequality, access and learning are actually changing.

What reconnection requires

Structured evidence only matters if it reaches the people and structures with the authority to act on it. The full paper sets out a tiered model of reconnection:

• At the frontline (micro level): timely, specific, psychologically safe structured insights that support reflective practice without overwhelming clinical teams.

• At the pathway and team level (meso level): aggregated themes mapped directly onto operational planning and pathway redesign.

• At board and system level (macro level): longitudinal trend analysis visible alongside finance, activity, clinical outcomes, and risk as a core indicator of system health.

Boards that embed experience evidence in formal governance — linked to risk registers, triangulated with staff and operational data, tracked over time — demonstrate that feedback is not merely collected but structurally decisive. This closes the loop: experience generates evidence, evidence informs action, action produces observable change, and that change becomes visible in the next cycle of experience data.

The policy asks: A national duty to learn from narrative

The next decade requires more than better local listening. It requires a nationally funded duty to learn from narrative. If the NHS is to prevent repeated cycles of harm, inquiry, litigation and recovery, patient, family, and staff narrative data must be preserved, protected, analysed, and routed into accountable governance.

This does not mean publishing raw personal material or weakening confidentiality. Privacy must be protected. But data protection should not become a shield behind which repeated warnings are obscured, diluted, or reduced to bland assurance language. The system needs a clearer distinction between protecting people and protecting institutions from scrutiny.

A future duty should require NHS organisations to preserve, analyse, audit and act on narrative evidence where it contains signals of risk, harm, culture, inequality, racism, discrimination, access failure, poor communication, unmet need, or missed learning. It should be introduced prospectively from a defined future date, with a time-limited implementation period that allows organisations to baseline their current position, disclose gaps, clean up data flows, and establish safe information governance rules. The purpose is not retrospective punishment; it is to create the conditions for honest learning.

This also needs central funding. Local Trusts cannot be expected to build national narrative intelligence infrastructure from exhausted budgets while managing operational pressure, workforce shortages, and recovery programmes. The money is already being spent downstream in litigation, compensation, crisis response, external reviews, inquiries, regulatory intervention, staff sickness, and reputational recovery. Government should prime the pump centrally, treating narrative intelligence as core safety and improvement infrastructure rather than an optional local analytics purchase.

The benefits extend beyond patient safety alone. Properly governed narrative intelligence can strengthen complaints resolution, reduce avoidable litigation, improve staff engagement, support psychological safety, identify inequalities, expose access barriers, inform pathway redesign, guide innovation, strengthen commissioning, improve regulatory assurance, and rebuild public trust. The test is not only whether people were heard, but what changed because they spoke.

The work ahead

Patient experience is not a peripheral metric. It is a holographic surface reflecting the complete system. Where listening is dispersed, learning drifts, where evidence is structured and reconnected, coherence emerges.

The next decade will require a nationally funded duty to learn from narrative: central government must prime the pump so that every NHS organisation is required and enabled to preserve, analyse, and act on the weak signals carried in patient, family, and staff voice.

Akumen's commitment is to this architecture. The work of the next decade is to move from survival to coherence — not as an aspiration, but as a design requirement. The full paper sets out the conditions, the evidence, and the direction. The question is whether the system will choose to build it.

The full paper — The Future of Patient Experience — is available from Akumen from 1st August. Contact: info@akumen.co.uk