Waiting list validation is not a data problem and it is not a technology problem. It is an outbound calling problem. That single distinction is the reason most NHS trusts are still running the slowest, most expensive version of it available, and it is the reason the manual model is now on a clock.
There is a structural explanation for why a trust can hold an accurate booking system and still carry a validation backlog it cannot clear. It has nothing to do with how well your access team performs. The work that closes the gap is sat in front of you every day, and the constraint is not effort. It is the number of patients a fixed team can actually reach by phone in a week.
Where the waiting list actually sits in 2026
The headline number is familiar, so the value here is the distribution behind it. As of March 2026, the elective waiting list in England stood at around 7.1 million pathways, covering roughly 6 million individual patients, the lowest level since August 2022 (Health & Protection, April 2026). The interim target of 65% of patients waiting less than 18 weeks was met for March 2026, and the share of patients waiting beyond 52 weeks fell close to the sub-1% target set for the same date.
What gets less attention is how that reduction was achieved. A significant part of it came from removing people who, for various reasons, should no longer have been on the list at all (The King’s Fund, December 2025). In other words, validation has already been doing a large amount of the heavy lifting in the recovery figures. It is not a back-office hygiene task. It is one of the few levers that moves the headline number without requiring an extra theatre list.
The gradient ahead makes that point harder to ignore. The medium term planning framework sets a national 18-week performance target of 70% for 2026/27, rising to the full 92% constitutional standard by 2028/29 (NHS Alliance briefing on the 2026-2029 framework). The same framework is explicit that waiting lists should reduce during 2026/27 by prioritising patients on clinical need and validating lists. Validation is now a named operational requirement, not a discretionary improvement project. Every trust COO in England is carrying this on their delivery plan whether they have resourced it properly or not.
For the access and booking lead reading this, none of the above is news. What may be useful is seeing it framed as a sequencing problem. The 70% target is a 2026/27 milestone. The validation work that supports it has to happen first, against a list that is still receiving close to 2 million new RTT pathways a month nationally. Validation is not a one-off cleanse. It is a recurring contact obligation that scales with the list.
This is a volume problem, not a performance problem
It is worth separating the cause here from anything that looks like a failure of your function. A validation team sized for steady-state booking activity cannot validate a list of 25,000 or 40,000 outpatient pathways at the pace the targets demand. The arithmetic does not close with the headcount most trusts hold.
Take a routine validation campaign across a few high-volume specialties. Orthopaedics, ophthalmology, ENT, gynaecology. Each patient needs a contact attempt, often more than one, then a structured response captured, then a PAS update, then a decision on whether the pathway continues, gets re-referred, or comes off the list. A band 3 administrator working a phone can complete a relatively small number of these in a day once you account for no-answers, callbacks, voicemails, and the patients who need two or three attempts before they pick up. Multiply the realistic daily completion rate by the size of the list and the timeline runs into months, not weeks.
That is the point at which agency and bank staff get pulled in to backfill, and it is also the point at which the model becomes both slow and expensive at the same time. The constraint was never the quality of the team. It was the ceiling on concurrent phone contact that any fixed headcount imposes.
Before the workflow, one number reframes what is possible at the contact stage, and it comes from a programme that has already been published in full.
What the NECU benchmark actually shows
The clearest published evidence on validation at scale comes from Scotland. NHS Scotland’s National Elective Coordination Unit ran a national waiting list validation programme that, in partnership with the digital provider DrDoctor, validated over 81,000 patients across eight health boards and more than 30 specialties in under six months (DrDoctor case study). The programme reported a 97% response rate, a 72% response rate within the first 24 hours, contact with 4.4 times more patients in a third of the time, and £3.57m in cost avoidance from digitising the process.
Those are strong numbers, and they make a real case for moving validation off pure manual calling. They also contain the detail that most summaries leave out, and it is the detail that matters for any trust planning its own programme.
NECU did not run text-only. The published service design sent an initial text message, and where no response was received, the team followed up with phone calls. Patients who only held a landline number were contacted by phone from the start (NHS Orkney NECU notice). The flagship digital validation programme in the UK could not complete the job by text alone. It needed voice to reach the patients the text did not.
That is not a footnote. It is the whole design problem in one line. A validation programme is only as good as its reach into the hardest-to-contact cohort, and on most elective lists that cohort is over-represented in exactly the specialties under the most pressure.
Why text-only validation leaves a gap
The patients least likely to respond to a survey link are often the ones who have waited longest. Older patients on orthopaedic and ophthalmology lists are less likely to hold a smartphone, less likely to engage with an SMS link, and more likely to rely on a landline. There is also a meaningful group with no mobile number on file at all, and a further group whose recorded mobile is out of date.
A text-first programme handles the responsive majority efficiently. It then leaves a residual cohort that is disproportionately made up of the patients a trust most needs to hear from, because their pathway has been open the longest and their clinical position is most likely to have changed. Validating the easy 70% and leaving the hard 30% does not clean the list. It cleans the part of the list that was least likely to be wrong in the first place.
This is why the choice is not really manual versus automated. It is about which channels a validation programme can actually run at scale. There are three models in practice, and they cover different parts of the list.
A manual team running phones reaches everyone in principle, including landline-only patients, but it is capped by concurrent calling capacity and leans on agency backfill to hit any meaningful timeline. A text-only automated programme is fast and cheap for the responsive cohort, but it structurally excludes the patients without a usable mobile and under-reaches the digitally excluded. Managed AI-voice outbound sits in the middle. It runs concurrent phone contact at a volume no fixed team can match, it reaches the landline-only and non-responding cohort that text misses, and it captures a structured response into the record on the call rather than relying on a patient to complete a form later.
That middle position is the defensible one, because it is the only model that combines the reach of phone with the scale of automation. It does not ask the trust to choose between contacting everyone and contacting them quickly.
The validation workflow, end to end
It helps to map where automation actually sits in the process, because the answer is not “all of it.” Validation runs as a sequence, and only one stage is genuinely a high-volume contact problem.
The first stage is list segmentation. The trust decides which pathways to validate, by specialty, by wait length, by clinical risk, and by data completeness. This is clinical and operational judgement and it stays with the team.
The second stage is outbound contact. This is the volume stage. It is the part that consumes band 3 time, pulls in agency staff, and sets the timeline for the whole programme. This is the stage where concurrent AI-voice calling changes the economics, because it removes the ceiling on how many patients can be reached in parallel. A text layer can run alongside it for the responsive cohort, with voice carrying the patients text does not reach.
The third stage is structured response capture. The patient confirms whether they still need the appointment, whether their condition has changed, and whether their details are correct. Capturing this in a structured form on the call, rather than as a free-text note to be processed later, is what lets the rest of the workflow run cleanly.
The fourth stage is the PAS update and the decision. Does the pathway continue, get re-referred, or come off the list. This is where clinical prioritisation and booking judgement live, and it stays firmly with the trust’s own team. The value of automating the contact stage is that it hands this team a clean, structured set of validated responses to act on, rather than a backlog of calls still to make.
Read that sequence back and the role of the team becomes clearer, not smaller. The validation team is the team. What changes is the composition of its week. The repetitive, capacity-capped contact work that currently sets the timeline gets absorbed in parallel, and the time recovered goes to the parts of validation that genuinely need a person: the clinical prioritisation, the re-referral decisions, the complex and sensitive cases, and the booking that follows. The alternative to a manual contact stage is not your own staff. It is agency backfill and a programme that runs in months instead of weeks.
The agency clock running underneath all of this
There is a national pressure that makes the manual contact model harder to sustain than it looks on a single business case. NHS England is driving agency spend down hard. The service cut roughly £1bn from agency spend in 2024/25, set a further 30% reduction target for 2025/26, and stated an aim to eliminate agency use across the NHS by the end of the current parliament, with possible legislation if progress stalls (NHS England letter, June 2025).
A validation model that depends on agency or heavy bank backfill to hit its timeline is therefore building on ground that is being deliberately removed. The independent picture on whether the crackdown is saving money is genuinely mixed, with some analysis showing bank costs running close to or above agency in places (Recruiter, May 2026). That nuance matters and it is worth being honest about. But the policy direction is not ambiguous. Trusts are being told to deliver more validation while leaning less on the temporary staffing that has historically made manual validation possible at pace.
This is the structural squeeze. More validation is mandated. The cheap way to surge manual contact capacity is being closed off. The only way that math resolves is by changing what does the contact, not by asking a smaller pool of staff to make more calls.
Where managed AI voice fits
This is the gap Auxilis was built to cover, on the channel that cannot be digitised away. The product, Jackie, is an AI voice system that holds natural two-way conversations with patients, captures structured information during the call, and writes it back into existing systems. It was proven first in NHS primary care, where the same underlying capability matters: reaching patients by phone, at volume, and turning the call into structured data the team can act on.
The primary care evidence is worth stating plainly because it is first-party and recent. Across live GP deployments, Jackie has absorbed 81% of inbound calls at one site, run a 91.1% call completion rate, handled more than 1,200 patient calls, and recorded zero outages. That is inbound, and waiting list validation is outbound, but the capability that decides whether validation works at scale is the same one. It is whether the system can reach the patient by phone, hold a real conversation, and record a clean structured response. Reaching the landline-only and the non-responding cohort by voice is precisely the part that text-only programmes cannot do and that manual teams can only do slowly.
Managed AI-voice outbound for validation is the logical extension of that capability. For a trust, the question to evaluate is narrow and concrete. Can a system contact the full validation cohort by phone in parallel, reach the patients text misses, capture a structured response that updates the PAS cleanly, and hand the clinical and booking decisions back to the team. That is the standard any validation approach should be measured against, manual or automated.
Frequently asked questions
Is waiting list validation mandatory for NHS trusts?
It is now a named requirement. The medium term planning framework for 2026-2029 states that waiting lists should reduce during 2026/27 by prioritising patients on clinical need and validating lists, alongside the 70% 18-week performance target for the year. Validation is part of how trusts are expected to deliver that target, not an optional add-on.
Can you validate a waiting list by text message alone?
Not completely. A text-first programme handles the responsive cohort efficiently, but it under-reaches patients without a usable mobile number, landline-only patients, and the digitally excluded. NHS Scotland’s national programme, often cited as the leading digital example, ran text first and then used phone follow-up for non-responders and landline-only patients. Voice was required to reach the patients text could not.
What is the difference between manual and AI-voice validation?
Both reach patients by phone. A manual team is capped by how many calls staff can make at once and usually needs agency or bank backfill to hit a campaign timeline. AI-voice outbound makes concurrent calls at a volume no fixed team can match, reaches the same landline and non-responding cohort, and captures a structured response on the call. The clinical prioritisation, re-referral, and booking decisions stay with the trust’s team in both models.
Does AI-voice validation replace the validation team?
No. The team remains. The contact stage is the part that gets absorbed in parallel, which is the stage that currently sets the timeline and pulls in agency staff. The time recovered goes to clinical prioritisation, re-referral decisions, complex cases, and booking. The alternative being removed is agency backfill, not the trust’s own staff.
Is AI voice safe and compliant for patient contact in the NHS?
Clinical safety and data governance are the right questions to ask of any system that contacts patients and writes to a clinical record. The relevant standards are clinical risk management under DCB0129, the Digital Technology Assessment Criteria, and NHS data protection obligations. Any vendor should be able to show how its system meets these before a pilot, and a trust should treat that evidence as a gating requirement.
How long does a validation programme take?
The timeline is set almost entirely by the contact stage. With a manual team, a large multi-specialty list runs into months. The Scotland programme validated over 81,000 patients across eight health boards in under six months using a digital-led model with phone follow-up. Removing the ceiling on concurrent contact is what compresses the timeline.
The fastest way to see whether this fits your list
The open question this post started with was why an accurate booking system can still carry a validation backlog. The answer is that validation is a contact problem, and contact is capped by how many patients you can reach by phone at once. Manual teams hit that ceiling. Text-only programmes step around it by leaving out the patients hardest to reach. The model that closes the gap is the one that runs phone contact at scale and reaches everyone.
The most direct way to judge whether managed AI-voice outbound fits your validation cohort is to see it handle a live call flow. The Auxilis demo runs in 20 minutes, on your existing telephony, with no hardware and no commitment. From there, a pilot lets you measure reach and structured-response capture against a defined slice of your own list before any wider decision. If your 2026/27 plan already has validation on it, that is the cleanest way to put a number on what a non-manual contact stage would actually do.