Introducing a new technology into a GP surgery is not the hard part. The hard part is everything that happens next.
Between 60 and 70 per cent of all organisational change initiatives fail, according to research published across multiple management studies. The most cited reason is not faulty technology. It is faulty implementation. Specifically, employee resistance and insufficient management support account for 72 per cent of transformation failures. The technology works. The change around it does not.
This is the conversation that tends to get skipped when AI voice systems arrive in primary care. Everyone focuses on whether the technology is ready. Far fewer ask whether the organisation is.
Before addressing healthcare specifically it is worth considering a parallel that most people in the UK have lived through personally.
Self-checkout machines made their UK debut in the early 2000s. The response was immediate and largely hostile. Customers complained they were slow, unreliable, impersonal and stressful. More than 236,000 Tesco customers signed a petition against the expansion of self-service tills. Booths removed them almost entirely in late 2023, citing customer preference. Morrisons admitted publicly that they had gone too far and too fast. Between a quarter and a third of shoppers still report that they dislike self-checkout.
And yet today, according to YouGov, 87 per cent of UK shoppers have used self-checkout. Larger retailers now dedicate up to 50 per cent of their tills to these machines. 74 per cent of consumers say they accept businesses using automated checkouts instead of human staff. The technology that provoked a national conversation about the end of human interaction is now simply part of how people shop.
What changed was not the technology. What changed was familiarity, design improvements driven by real-world feedback, and the gradual recognition by shoppers that the alternative, long queues, limited opening hours, inconsistent service, was not actually better. The 10 per cent who still dislike self-checkout are real and their preference matters. But the 90 per cent who now use it without a second thought were once in the same camp as the petitioners.
AI voice systems in GP surgeries are at the beginning of exactly this curve. The complaints are real. The resistance is predictable. And the direction of travel, for patients who get used to a system that answers immediately, never keeps them on hold for an hour, and handles their booking in five minutes rather than twenty, is equally predictable.
The supermarket analogy only goes so far, because the NHS is not a supermarket. The stakes are higher, the regulatory environment is more complex, the workforce is under greater pressure, and the patients using the system are often more vulnerable than a shopper choosing between tills.
This means that change management in primary care AI is not just good practice. It is a clinical responsibility.
Research consistently shows that 30 per cent of employee adoption failure is directly attributable to insufficient training. Yet organisations globally allocate only around 10 per cent of transformation budgets to change management. The gap between those two figures is where most implementations go wrong.
When EMMA goes into a GP surgery, the technology is ready. But whether it works well for patients and staff depends enormously on what happens in the practice around it. A practice where the manager has communicated clearly with staff about what the system does, why it is being introduced, how it handles patient calls and what to do when it cannot help is a practice where EMMA works well from day one. A practice where it has been switched on without preparation, where receptionists feel threatened by it, where patients have not been given any information about what to expect, is a practice that will generate complaints regardless of how good the technology is.
QuantumLoopAI supports its surgery partners through providing digital tools, marketing materials, patient communications and training resources to help practices introduce EMMA effectively. That support covers the technology side of the transition thoroughly.
But the honest truth about change management in any organisation, and particularly in the NHS, is that no supplier can do it for you. The practice manager, the GP lead, the business manager, whoever holds operational responsibility for how the surgery runs, has to own the implementation. They are the ones who know their staff, their patient population, their particular pressures and sensitivities. They are the ones who can make the difference between a rollout that generates resistance and one that generates confidence.
This is not a criticism of surgeries that have struggled with implementation. It is a recognition that asking a busy NHS practice to manage a significant technology change, on top of everything else, without adequate time, resource or change management support, is genuinely hard. It is one of the most challenging things a practice manager can be asked to do. Acknowledging that difficulty honestly is more useful than pretending the technology will handle it on its own.
One of the most common mistakes in any technology implementation is failing to have an honest conversation with staff before the system goes live. In primary care this tends to manifest in two ways.
The first is receptionists feeling that the AI is a judgment on their performance, a signal that they are being replaced or that their work was not good enough. As explored elsewhere in this series, the opposite is true. EMMA handles routine call volume so that receptionists can do the work that actually requires a person. But that message has to be communicated, clearly and repeatedly, before resentment takes hold.
The second is the misconception that once the AI is in place, the workload disappears. It does not. What changes is the nature of the work. Patient submissions and requests that EMMA handles will surface in a different form for staff to manage and act on. There will be a period, particularly in the first months of any deployment, where it feels like more work rather than less, because staff are learning a new system while still managing the transition. This is normal. It is temporary. And it is significantly less disruptive if staff have been told to expect it in advance.
Training completion correlates directly with higher system usage and adoption rates after rollout. Practices that invest in proper staff training before and during implementation consistently see better outcomes than those that do not. This is not a complex insight. It is one that is regularly ignored in the pressure to get a new system live as quickly as possible.
Patients also need change management, even if nobody usually calls it that.
When a patient rings their GP surgery and encounters an AI voice system for the first time without warning, their first response is almost always negative. Not because the system is bad but because it is unexpected. Surprise in a healthcare context, when someone is already worried about a symptom or struggling to get through, generates anxiety that becomes frustration that becomes a complaint.
But there is a deeper problem than surprise. Most patients have no idea why the change has been made. They do not know that their surgery receives hundreds of calls every morning, that receptionists were spending the majority of their day answering the phone rather than managing care, or that the 8am scramble meant many patients could not get through at all. They just know something is different and they did not ask for it.
Patients who understand why a change has happened are far more likely to engage with it constructively. They do not need a technical briefing. They need a simple, honest explanation that connects the change to something they already care about.
That explanation is not complicated. EMMA handles routine calls so the surgery can focus on actual care. It means your call is answered immediately rather than after an hour on hold. It means the receptionist has time to follow up on your referral, chase your results and have a proper conversation when you need one. It means time and money that was going on call handling infrastructure goes toward clinical care instead. This tool does not exist to make things harder. It exists to fix something that was already broken and already making things harder every single day.
The supermarkets that managed the self-checkout transition best were not the ones that installed machines and hoped customers would adapt. They were the ones that explained the benefit clearly and kept a human option available. The NHS cannot afford the reversal that several retailers had to make when they moved too fast without communication. Getting patients to understand the bigger picture is not optional. It is the difference between a technology that earns trust over time and one that becomes a controversy.
The 10 per cent of patients who will always prefer speaking to a person are real and their needs must be respected with a clear fallback at every point. But the majority, given honest communication and time to get familiar, will get there. They got there with online banking. They got there with self-checkout. They will get there with this too.
The NHS 10 Year Health Plan commits the health service to a fundamental shift toward digital-first primary care. That transformation will require thousands of individual change management processes across GP surgeries, primary care networks and integrated care systems across England. Most of them will be imperfect. Some will be difficult. A few will fail.
The ones that succeed will not be the ones with the best technology. They will be the ones where someone took change management seriously: communicated clearly, trained their staff properly, prepared their patients, and gave the transition enough time and support to work. Technology is the easy part. People are the hard part. They always have been.
Sources: