Summary
Build it on documentation minutes, not on clinic capacity. The largest published NHS evaluation, across nine London sites and more than 17,000 encounters, found 23.5 percent more time in direct patient interaction and appointment length down 8.2 percent. A study of 1,800 clinicians across five academic centres found 16 minutes saved per eight hours of patient care. Both are real and neither creates an extra appointment, which is the claim most business cases are wrongly built on.
Most AI scribe business cases fail the same way. They multiply a per-consultation time saving by the annual consultation count, convert the hours into appointments and present a number that a finance director recognises immediately as not real, because the freed minutes arrive in two-minute pieces spread across a clinic rather than in a block anyone can book into.
The published evidence is better than that approach and supports a smaller, more defensible claim. It is worth building the case on what was actually measured, because the measured effects are genuine and a case that survives scrutiny gets funded twice.
Everything below is a published figure with its study named. Convert none of it into revenue without saying out loud which conversion you are assuming.
See it in your clinicOur price is published, so the cost side of your case takes four seconds rather than a procurement cycle.
Request a demo| Study | Scale | Finding |
|---|---|---|
| NHS England sponsored, led by GOSH DRIVE | 9 London sites, 17,000+ encounters | 23.5 percent more direct patient interaction time, appointment length down 8.2 percent |
| Five academic medical centres, 2023 to 2025 | 1,800 clinicians | 16 minutes of documentation saved per 8 hours of patient care |
| General practice, before and after | 535 consultations | 42.7 seconds less documentation per consultation, total consultation time unchanged |
| The category's randomised trial | 238 physicians, ~72,000 encounters | 9.5 percent less note-writing time, about 41 seconds, no fall in after-hours record time |
| Outpatient mixed methods trial | 100 clinicians, 7,499 consultations | 58 percent of drafts accepted unedited, 47 percent of staff reported hallucinations |
The honest number is small and it is still worth having
Sixteen minutes per eight hours is two minutes an hour. Forty-one seconds a note across a thirty patient list is about twenty minutes. Forty-two point seven seconds per consultation in general practice, with total consultation time unchanged, means the time came out of the documentation rather than out of the appointment.
None of those numbers buy you an extra clinic. They buy the difference between finishing at six and finishing at half past five, which is a real thing to buy and a terrible thing to put in a spreadsheet as revenue.
The GOSH DRIVE figures are the most useful for a case because they measure something a clinician recognises. Twenty-three and a half percent more time in direct patient interaction is not time saved, it is time moved from a screen to a person inside the same appointment. That is a quality argument with a number attached, which is rarer and more persuasive than a productivity one.
The three claims that get a case rejected
That saved minutes become appointments. They do not, unless you actually shorten the appointment length and book into the gap, which is a separate decision with its own clinical consequences. The general practice study is explicit that documentation time fell and total consultation time did not.
That the scribe pays for itself through coding. In the United States that argument has real support, because the note is the evidence for the E/M level. In UK private practice and in the NHS there is no equivalent mechanism, and importing the US coding argument into a British business case is the fastest way to lose the room.
That the comparison is against a human scribe. A human scribe costs tens of thousands a year and almost no UK practice was ever going to employ one, so the comparison flatters every product in the category and answers a question nobody asked. The real alternative is the clinician typing the note after clinic.
What to measure before you buy
Two weeks of baseline, and it is the only part that cannot be recovered later. Minutes spent documenting outside session hours, time from consultation to a signed note, the share of letters leaving within a week, and the size of the unworked recall or results queue.
Then cost the whole thing rather than the licence. Subscription, the implementation and training you will actually need, the clinical safety and information governance work, and the fortnight where two systems run in parallel. Our guide to the clinical safety standards sets out the work that is easy to leave out of a budget.
State the conversion assumption explicitly. If the case says forty minutes a day returns to the clinician, say what you expect them to do with it, and whether that is finishing earlier, seeing one more patient, or clearing the letters that currently leave late. All three are defensible. Leaving it unsaid is not.
The largest number is usually outside the scribe
Every figure on this page measures the note. In most clinics the note was never the expensive part of the day. The enquiry that went unreturned, the result nobody chased, the recall list nobody worked and the letter that left a week late cost more per month than any subscription in this market, and they are paid for in people.
That is why a business case built only on documentation minutes tends to look marginal even when the product is good. The product is answering a real but small question.
If the case is struggling to clear the bar on documentation alone, the honest conclusion may be that the scribe is the wrong purchase rather than that the numbers need massaging. Our piece on the front desk and the consultation makes that argument in full.
Where wai fits a case
wai is priced on the work around the consultation rather than on the note, so the case is built differently. The enquiry is answered, the booking is taken, the note is composed, the letter is drafted for approval and the result is chased, on one record, at a figure published on our pricing page.
That makes the comparison a staffing one rather than a licence one, which is a harder case to write and an easier one to defend, because the costs it displaces are on a payroll rather than in a benchmark.
We are not the cheapest and a clinic whose binding constraint is the monthly software bill should buy a scribe and stop reading here. Clinicians still decide what enters the record.
Bring your own numbers to a call. Fifteen minutes, on your clinic rather than on a benchmark.
Request a demoFrequently asked questions
How much time does an AI scribe actually save?
Less than the marketing and more than nothing. A study of 1,800 clinicians across five academic centres found 16 minutes saved per eight hours of patient care. A general practice study of 535 consultations found documentation time down 42.7 seconds per consultation with total consultation time unchanged. The category's only randomised trial found about 41 seconds a note.
Can we convert the time saved into extra appointments?
Not without deciding to, separately. The freed minutes arrive in short pieces across a clinic rather than as a bookable block, and the general practice evidence shows documentation time falling while total consultation time stayed the same. If the case assumes extra appointments, say so explicitly and say what changes in the diary to create them.
What did the NHS evaluation find?
An NHS England sponsored study led by GOSH DRIVE across nine London sites, covering more than 17,000 patient encounters, found clinicians using AI scribe technology spent 23.5 percent more time in direct patient interaction, with overall appointment length down 8.2 percent. That is time moved from a screen to a person rather than time removed from the day.
Does the coding argument work in a UK business case?
No. The claim that a better note pays for itself through coding rests on US payer mechanics, where the note is the evidence for the E/M level. Neither UK private practice nor the NHS has an equivalent, so importing that argument tends to lose the room rather than win it.
What should we measure before buying?
Two weeks of baseline, because it cannot be recovered later. Documentation minutes outside session hours, time from consultation to signed note, the share of letters leaving within a week, and the size of the unworked results or recall queue. Then cost the whole deployment rather than the licence.
What does an AI scribe deployment cost beyond the subscription?
Implementation and training, the clinical safety work under DCB0160 including a named Clinical Safety Officer, a data protection impact assessment, and the fortnight where two ways of working run in parallel. Suppliers rarely price these and finance directors always find them.
Why does the business case often look marginal?
Because every published figure measures the note, and in most clinics the note was never the expensive part. The unreturned enquiry, the unchased result, the unworked recall list and the late letter cost more per month than any subscription in this market. A case built on documentation minutes alone is answering a real but small question.
Sources
- Ambient scribe in general practice, a multi-perspective before-after longitudinal mixed-methods study
- Performance, acceptability and impact of ambient listening scribe technology in an outpatient context, BMC Health Services Research, January 2026
- Large AI scribe study finds modest time savings and inconsistent use, STAT, April 2026
- Evaluating an artificial intelligence scribe for clinical documentation, Digital Health, 2025
Build the case on your own numbers
Fifteen minutes on your workflows. If it is a fit we reply within 24 hours.
Partner with us
See it on your own patients, then build it with us
Thirty minutes on your own patients, doctors and rules, in your market. We reply within one working day, and the clinics that partner with us early shape what we build next.
See what wai does between the visits