How PCNs Can Roll Out Digital Triage Across Member Practices

At 8am, one practice has switched on online access while the neighbouring practices are still handling demand through the telephone. By mid-morning, the first practice has a queue of digital submissions, the second has a full call queue, and nobody has a reliable PCN-wide view of which requests are waiting, who owns them, or where clinical capacity remains.

That isn't a software problem. It's a network design problem. Digital triage changes the route into general practice, so a PCN needs shared rules for intake, prioritisation, booking, escalation, staffing and measurement across every member practice. This guide sets out how PCNs can roll out digital triage across member practices without just moving the bottleneck from the telephone to an inbox.

The problem a PCN digital lead walks into on Monday morning

The Monday morning queue exposes every weakness in a fragmented access model. Patients call the practice with the shortest hold time, submit online requests wherever the form is easiest to find, or walk into reception because they aren't sure which route applies. Reception teams make local decisions, clinicians pick up work at different points in the day, and practice managers often can't see the total demand arriving elsewhere in the network.

One practice may introduce an online consultation pathway and reduce pressure at its front desk. That doesn't mean PCN demand has reduced. Patients who can't use the digital route, don't understand the eligibility rules, or find the online access point closed may shift to another practice's telephone line. The PCN has improved one queue while leaving the underlying distribution problem untouched.

Healthwatch's evidence review describes the NHS England total triage model as an approach where every patient contact is assessed before an appointment is made, either by telephone or through online consultation. The review also says online consultation was expected to create further efficiency gains, but its evidence shows why digitisation alone isn't enough. In one GP access dataset, face-to-face GP appointments rose from approximately 9,125 in May 2021 to 21,570 in October 2021, a 236% increase, while phone consultations fell by 9% over the same period, as reported in the Healthwatch evidence review of GP access.

For patients, the access experience is shaped by the whole network, not by the procurement decision of one practice. PCNs should therefore standardise three things from the start:

• Demand capture: Every practice records requests through an agreed set of channels and categories.
• Capacity ownership: Each request has a named destination, escalation route and accountable team.
• Network analytics: Practice-level activity is combined into a PCN view so leaders can identify migration, backlog and uneven workload.

A useful patient engagement perspective is available in this healthcare patient engagement guide, particularly for thinking about how patients receive instructions and move between access channels. It shouldn't replace clinical or operational design, but it can help PCNs test whether their communication is clear enough for patients to use the intended route.

Five models that PCNs should distinguish

The term “digital triage” covers several materially different operating models. A board should name the model it is approving rather than treating every online form or automated workflow as equivalent.

1. Manual reception triage keeps the first decision with trained reception staff. It can provide a consistent front door when staff have clear scripts, escalation rules and access to clinical support, but it remains dependent on human availability and confidence.

2. Manual GP-led triage sends requests to a clinician for review before booking. It can support complex prioritisation and gives the practice direct clinical oversight, but it places the triage workload on GPs or other clinicians.

3. Form-based online consultation captures information digitally before a human reviews it. This improves the quality and consistency of information arriving at the practice, but the practice still needs someone to assess each submission.

4. AI-assisted triage helps a human clinician or staff member interpret and route a request. That is a genuine improvement where it reduces searching, repetition or administrative effort. The important distinction is that a human still makes the decision.

5. Autonomous AI triage allows the system to assess and route the request according to its configured clinical and operational rules. In this model, the system makes the defined triage decision, with escalation and oversight governed by the practice's safety model.

The first four models digitise, accelerate or reorganise triage. The fifth is the model that removes the manual triage step. That doesn't make the earlier approaches useless. It means PCNs need to be honest about the workload each model leaves in place.

A 2024 UK primary care case report described a single-practice PCN where call volumes fell by around 60%, from approximately 200 calls by 8am on Monday to 80, while approximately 82% of appointments were booked through the triage tool. The same report says NHS England guidance asked PCNs to monitor telephony data across their service and combine practice-level data into a PCN-wide picture. Its lesson is operational, not promotional: the PCN has to measure the route demand takes after rollout, not just whether a tool is switched on. The case report on digital triage and PCN call volumes provides that context.

Building the business case your PCN board will actually sign off

A PCN board won't approve a credible rollout because a supplier promises a shorter queue. It needs to understand the starting workload, the proposed operating model, the safety controls and the decision points that will stop a weak pilot from expanding.

Start with a baseline for each practice. Record how requests arrive, who handles them, how many are converted into appointments, where urgent work is escalated and which teams carry the administrative burden. Don't combine GP-led triage time with general operational hours. If the baseline distinguishes them, the business case should preserve that distinction.

Next, model the expected change in demand distribution, not an invented saving. Ask what happens if more requests arrive online, whether appointments remain practice-owned, and where the work goes when a practice reaches capacity. NHS England's total-triage blueprint supports a staged approach, with online consultation access available across practices and the option of a virtual hub where clinicians work across multiple practices when technical infrastructure, record sharing and smart-card access are in place. The NHS England total-triage implementation blueprint is the relevant reference for that design work.

Your board paper should answer four questions plainly:

• Who owns the inbox? Name the team and the clinical escalation route.
• What happens when appointments are full? Define routing, safety-netting and capacity escalation before go-live.
• How is safety assured? Set out the clinical safety documentation, incident process and review cadence.
• What happens if one practice withdraws? Document the effect on shared staffing, routing and patient communication.

Treat ROI as a method, not a promise. Use your own demand, staffing and capacity inputs, then set pilot exit criteria that compare workload distribution, safety events, patient experience and equity across practices.

Governance, clinical safety and procurement criteria

The PCN should lead the assurance process rather than accepting a supplier's compliance statement as the whole safety case. A multi-practice deployment creates shared dependencies, so each member practice needs to understand its responsibilities even when procurement is organised centrally.

The evidence pack to request

Ask for the documentation that supports both product and organisational assurance. That includes DCB0129 documentation for the manufacturer's clinical safety work and DCB0160 documentation for the deploying organisation's clinical safety management. The PCN's Clinical Safety Officer should use these documents to build or review the hazard log, define local controls and record approval conditions.

The information governance pack should cover the PCN's processing arrangements, access controls, retention, data flows and DPIA requirements. A single PCN procurement doesn't remove the need to understand how each practice participates in the processing and clinical workflow.

Procurement questions should be specific:

• What is the regulatory status of the clinical engine?
• What evidence supports its intended use?
• How are triage decisions represented and audited?
• What information is written into the booking record?
• How does the system integrate with the practice's clinical system?
• Which organisation owns configuration, incident review and change control?
• What happens when a practice has different opening hours, staffing or escalation rules?

GP Triage is one example of the autonomous model. It uses an Infermedica Class 2B medical device, with UKCA/MHRA, DCB0129 and DCB0160 documentation, supports adult and paediatric presentations, integrates with the major UK clinical systems, and pushes a triage summary into the booking record. Those are procurement facts to verify against the PCN's intended workflow, not reasons to skip comparative assurance.

Its stated safety wording should remain precise: “to date, zero recorded clinical safety incidents across approximately two million triages” and “around 97% concordance with GP decision-making”. Those statements describe reported experience and concordance. They aren't guarantees, so the PCN still needs local configuration, monitoring and escalation.

Integrating with practice clinical systems and NHS Spine

The integration architecture determines whether digital triage reduces handling or creates another reconciliation task. PCNs should approve the workflow on paper before they approve the technical deployment.

The first decision is appointment ownership. A PCN can keep each practice's slots sovereign, create agreed cross-practice capacity, or use a virtual hub with clinicians working across sites. None of these options is automatically right. The choice depends on the network's governance, staffing, record-sharing arrangements and ability to provide smart-card access for clinicians working across practices.

The second decision is what happens after triage. The output should create a usable, auditable summary in the booking record and the practice's clinical record, according to the agreed workflow. A PCN shouldn't accept vague assurances about “integration”. Ask to see the fields, the destination, the audit trail and the failure process when the write-back doesn't complete.

NHS Spine integration and patient identity verification also need to be treated as part of the design, not as a late technical detail. If a virtual hub handles requests for multiple practices, the PCN must know how identity is confirmed, how access is restricted and how the receiving clinician sees the right information for the right patient.

Capacity rules under continuous access

The 2025/26 GP contract change requires online consultation tools to remain open throughout core hours for non-urgent requests, medication queries and administrative requests, with safeguards to reduce the risk of urgent clinical requests being submitted incorrectly. The NHS England guidance on the 2025/26 GP contract changes makes the operational implication clear: access hours and clinical capacity have to be planned together.

A PCN should decide whether submissions are routed to the patient's registered practice, a central hub or a defined shared service. It should also define what happens when a practice pauses local capacity, how that status is communicated, and who reviews urgent or misrouted requests. If those rules aren't written before go-live, reception teams will create them under pressure.

Multi-channel intake and training the people who actually use it

A digital front door fails when the PCN treats the website as the whole access model. Patients will still use the telephone and reception, and staff need a consistent way to capture requests without making digital confidence a condition of receiving care.

Configure the channels together. Web access can capture requests directly. Telephone teams can record information using the same triage logic where patients can't submit online. Reception staff can provide assisted entry or use an agreed handover route for people who need help. The objective is not to force every patient into one channel. It is to ensure that every channel feeds the same routing, urgency and capacity rules.

Reception training needs more than a demonstration of the interface. Staff should know what to say when a patient can't use digital access, how to recognise a request that needs immediate clinical escalation, what information they can record, and when they must hand over rather than resolve the issue themselves. Clinical teams need protected time to understand how the first-contact conversation changes when information arrives before review.

Use evidence without overstating it

The approved GP Triage examples show why PCNs must label hours carefully. Langton Medical Group, with approximately 14,000 patients across 3 sites, reports around 30 hours of GP-led triage removed per week, to date. Swanscombe Health Centre, with approximately 37,000 patients, reports around 422+ operational hours returned in the first four weeks, with around 5,000 appointments booked autonomously in that period, to date. These are reported reference figures, not a forecast for every PCN, and the Swanscombe figure is operational rather than GP-specific. The GP Triage reference examples should be read in that context.

Choose pilot sites fairly. Don't select only the practice with the strongest digital confidence and easiest demand profile. Include variation in size, staffing, patient access needs and current workflow maturity, then document why each site was chosen.

Use a defined 6 to 8 week evaluation window. Agree the go or no-go criteria before the first patient submits a request:

• Workload redistribution: Has work moved to the intended route, or has it appeared in another practice's queue?
• Safety: Have incidents, near misses, misroutes or escalation failures been identified and reviewed?
• Patient experience: Can patients understand the access routes and complete them without avoidable friction?
• Equity: Are patients who need telephone or reception support still able to access the same safe triage pathway?
• Operational control: Can practice managers see outstanding work, capacity status and ownership?

One practice's success can still overwhelm its neighbour if the PCN has no shared routing rules. Agree escalation thresholds, define how practices share activity data, and give each site a documented way to pause local capacity without transferring responsibility. NHS England's blueprint treats PCN support as part of the implementation model, not an optional add-on. That means the network should own the SOPs, training, review and cross-practice coordination.

The PCN-level KPIs that prove digital triage is working

Submission volume measures adoption. It does not prove that a PCN has improved access or reduced pressure. NHS England reported approximately 6.5 million online submissions in September 2025, up around 49.9% year on year, and approximately 8.6 million in February 2026, up around 85% year on year, as noted in the contract-change guidance cited earlier. Treat these figures as demand signals, then test whether the network has redistributed work safely and fairly.

The BMJ analysis of online consultation use reported approximately 108.7 submissions per 1,000 registered patients in September 2025, alongside substantial geographic variation in usage across England. Compare member practices using identical definitions. Read differences alongside population needs, demographics, promotion and non-digital access. A high submission rate may indicate better access, greater need or a routing problem.

Metric	Practice level	PCN level	Source
Demand volume	Requests by channel, day and category	Demand share and redistribution between practices	Practice activity data and NHS England access infrastructure
Response pathway	Time to first response and route selected	Escalation pattern by site and service	Triage and booking records
Resolution	Share resolved without face-to-face review	Variation in resolution between member practices	Triage outcome data
Access friction	Abandonment and incomplete submissions	Differences by practice and channel	Channel analytics
Safety	Incidents, near misses and misroutes	Shared trend review and corrective actions	Local safety reporting
Capacity	Outstanding work and available slots	Cross-practice pressure and ownership	Booking and telephony data

Report the measures at practice and PCN level each review cycle. A network average can hide one practice carrying the queue while another retains unused capacity. Assign an owner to every red measure, record the corrective action, and set a review date. The board needs decisions, not a dashboard catalogue.

NHS England Digital's access dashboard supports views across practices, PCNs, ICBs, regions and the national system, including scheduled activity and cloud-telephony volumes. Use it where possible, then add fields for triage outcome, ownership, escalation and unresolved work. Data should show where the network must alter routing, staffing or shared capacity. A broader approach to using data to improve patient outcomes with analytics also starts with converting measures into operational decisions.

Risks that derail a PCN rollout and how to mitigate them

A PCN can open an online route on Monday and discover a hidden queue by Friday. Continuous access increases the flow of requests, but capacity, ownership and escalation still need explicit design. If those arrangements remain local and informal, the telephone bottleneck becomes less visible rather than less serious.

Continuous access without capacity ownership

The 2025/26 contract position makes online access availability a planning issue. Rising online demand means a PCN cannot ask each practice to absorb whatever arrives at its front door. The network needs a shared operating model before launch.

Set a schedule for reviewing requests, name the responsible team, define the maximum period before escalation, and agree who owns work when a practice reaches capacity. A practice should be able to pause or redirect local capacity through a documented process. Patients and neighbouring practices should not have to discover a change through unanswered requests or sudden appointment scarcity.

Treat capacity as a PCN resource where the operating model supports it. Record which practice owns each request, who can reassign it, and how unresolved work is reported at the next operational review.

Capacity disputes between member practices

A virtual hub can improve flexibility, but it also creates accountability questions. Who books the appointment? Which practice funds the clinician? Who follows up an incomplete request? Which clinical record receives the summary? The PCN board should answer each question in the SOP and service agreement before the pilot begins.

Use a named clinical lead for each site, a network-level operational owner and a monthly PCN safety review. Keep appointment sovereignty where shared booking would create unacceptable ambiguity. Centralise only after record sharing, smart-card access, clinical responsibility and information governance are ready.

The service agreement should also define workload transfer. If one practice accepts another practice's requests, specify the receiving team, funding arrangement, response standard and responsibility for follow-up. Without those rules, redistribution becomes a dispute rather than usable network capacity.

Safety drift from inconsistent routing

Different practices often use different definitions for urgent, routine, administrative and medication requests. Patients experience that as inconsistency, while staff experience it as rework and repeated clinical review.

Create one network routing matrix. Include urgent escalation, same-day capacity, routine review, administrative handling and requests that need human intervention. Record who can change the rules, what evidence is required, and how changes receive clinical approval. Autonomous triage can be configured to a practice's safety model, but configuration must be controlled across the PCN instead of altered informally at individual sites.

For GP Triage specifically, the approved safety wording is “to date, zero recorded clinical safety incidents across approximately two million triages” and “around 97% concordance with GP decision-making”. The PCN should still validate configured pathways locally, monitor incidents and maintain its own DCB0160 responsibilities.

Digital exclusion and channel failure

Digital-first access can favour patients who are confident online and disadvantage patients who need telephone or face-to-face help. Retain assisted routes, and train reception teams to use the same underlying workflow. A separate low-visibility queue makes access performance difficult to assess and increases the risk of inconsistent routing.

Monitor abandonment, incomplete submissions and the mix of telephone, web and reception capture by practice. If one practice shows a markedly different pattern, investigate the access journey before concluding that patients have lower digital demand. Check language needs, accessibility, staff prompts and whether callers are being offered the same clinical route.

A practical 90-day sequence

A defensible rollout can follow this sequence:

• Weeks 1 to 2: Secure the PCN board mandate, appoint the network operational owner and involve the Clinical Safety Officer.
• Weeks 3 to 6: Complete procurement, review DCB0129 and DCB0160 documentation, and complete the DPIA and information governance work across participating practices.
• Weeks 7 to 10: Configure the pilot workflow, appointment rules, escalation thresholds, multi-channel scripts and staff training at the first sites.
• Weeks 11 to 12: Go live with daily operational and safety review. Record demand movement, unresolved work, incidents and patient access problems.
• Following months: Scale in stages, reviewing a shared PCN dashboard monthly and stopping expansion when the agreed exit criteria are not met.

Board approval should depend on operational readiness, not on a software demonstration. Before each expansion, confirm that every participating practice has a named owner, agreed routing rules, usable capacity data, trained staff and a tested escalation route.

Continuous online access works only when the PCN redesigns triage and capacity ownership alongside the channel. Without that redesign, the old bottleneck returns through a new interface.

Book a working demonstration with GP Triage to examine how its autonomous AI triage and booking platform could fit shared PCN SOPs, clinical safety governance, multi-channel intake and integration decisions across member practices. Test the workflow against your own capacity rules, escalation thresholds and pilot criteria.