Key takeaways
- Mystery patient audits test access and information, never clinical judgement, and must not take real appointments from patients.
- Tell staff an audit period is coming, explain its purpose is improvement and set rules for any recordings before you start.
- Score each contact on a simple 0-2 scale across five areas so results are comparable between weeks and staff.
- Use the national GP Patient Survey for context, but your own audit shows where and when the friction happens.
What the audit is for
A mystery patient audit uses planned, scripted contacts to see your practice the way a patient does: ringing at 8am, using the online form, asking at the desk or sending a website message. It tests access, information and manner. It does not test clinical decisions, and it should never use invented symptoms that take a clinician's time or a slot a real patient needed.
National data give context. In the 2026 GP Patient Survey, 56.9% of patients said it was easy to get through to their practice by phone, up from 52.9% in 2025 and 49.7% in 2024[1]. Only 56.9% were offered a choice of time or day for their appointment[1]. Those averages hide large differences between practices and between hours of the day, which is what an audit uncovers.
Source: Ipsos (GP Patient Survey provider for NHS England), 2026[1]
Show the data
| Period | Value |
|---|---|
| 2024 | 49.7% |
| 2025 | 52.9% |
| 2026 | 56.9% |
Decide what you are testing
Pick two or three questions, not ten. Good audit questions are specific enough to change something.
- Can a caller find out how to make a routine request within a few minutes at peak time?
- Do the phone message, website and reception give the same instructions?
- Are queue position and call-back options working as intended? NHS England promoted both as standard digital telephony features[2].
- For private clinics: can a new self-pay patient get a price and a date in one contact? Among people who have gone private, 28% say they did so because booking is easier[3].
- Do staff offer an interpreter, an alternative format or a quiet time when asked?
Dental practices and private clinics
The method works the same way outside general practice, but the questions shift. In an NHS dental practice, the most revealing scenarios are a new patient asking whether the practice is taking on NHS adults or children, and a caller in mild discomfort asking how soon they can be seen. In a private clinic, the commercial journey matters most: how quickly an enquiry gets a named clinician, a price, a date and a confirmation in writing.
Private clinics should also audit their web enquiry forms and social media messages, since many self-pay patients never phone at all. Record how long a reply takes and whether it answers the question that was asked.
Planning the audit
Write scenarios
Four to six admin scenarios that a real patient would bring, each with a short script and a clear end point before any booking is confirmed.
Set the sample
Spread contacts across days and times, weighting towards peaks. Twenty to forty contacts over two to four weeks is enough for a small practice to see patterns.
Choose the auditors
Patient participation group members, staff from a neighbouring practice or an external provider. Avoid anyone the team would recognise by voice.
Agree the rules
No real patient identities, no clinical symptoms, end the call before a slot is held, and cancel at once if a booking is made by mistake.
Prepare the scorecard
Use the same scoring sheet for every contact and record date, time, channel and wait.
Consent, staff and data protection
Mystery audits involve your staff's personal data, especially if calls are recorded. Tell the team in advance that an audit will run during a stated period, explain that it is about systems rather than individuals, and say how results will be shared. Covert audits used for discipline damage trust and are hard to justify.
You need a lawful basis for any recording or note that identifies a member of staff. Private clinics often rely on legitimate interests, but UK GDPR says legitimate interests cannot be used by public authorities for processing in the performance of their tasks[4]. NHS GP practices are public authorities for their NHS work, so ask your data protection officer which basis applies and record the decision.
- Update the staff privacy notice to mention quality audits.
- Keep recordings only as long as needed to score them, then delete.
- Report results by scenario and time slot, not by named individual.
- Never use a real patient's name, date of birth or NHS number in a scenario.
Scenarios that work
| Scenario | What it tests | A good result looks like |
|---|---|---|
| Routine appointment request at 8.05am | Queue, triage route, online alternative | Clear route explained; wait time given; no need to call back |
| Repeat prescription question | Signposting to the right channel | Correct method explained in one go |
| Asking how to register a new family member | Information accuracy | Accurate steps; form or link offered |
| Request for an interpreter at the next visit | Inclusion and recording | Staff know how to book and note it |
| Self-pay price and availability enquiry (private) | Price clarity and booking ease | Price, inclusions and next date given |
| Website message out of hours | Response time and content | Acknowledged and answered within the stated time |
Write each scenario as a short brief rather than a word-for-word script: who the caller is, what they want, one detail that makes the request realistic and the point at which they politely end the contact. Auditors who read from a script sound unnatural, and staff respond to that. Give them a fallback line for when they are asked for identifying details, such as saying they will call back with their NHS number to hand.
A simple scoring framework
Score each contact from 0 to 2 in five areas: 0 means not met, 1 partly met, 2 fully met. A maximum of 10 per contact keeps scoring fast and makes trends easy to read.
- Reach: did the patient get through or get a reply, and how long did it take?
- Accuracy: was the information correct and consistent with the website and phone message?
- Resolution: did the contact end with a clear next step?
- Manner: was the patient treated with courtesy, without being rushed or blamed?
- Inclusion: were needs such as language, access or anxiety picked up and handled?
Record the wait time separately as a number. It is the measure most likely to vary by hour and the one most often fixed by changing rotas rather than people.
Reading the results honestly
Small samples tell you where to look, not what is proven. Before drawing conclusions, line the scores up against your call statistics and complaints for the same weeks. If mystery calls at 8am score poorly and your phone data show a spike in abandoned calls at the same time, you have a finding. If a single low score stands alone, treat it as a prompt to listen to more calls.
- Group scores by hour of day and channel before looking at anything else.
- Look at the spread as well as the average; one very poor contact matters more to a patient than a slightly lower mean.
- Separate system problems, such as a queue message or form design, from training needs.
- Note what went well and say so to the team; good practice spreads faster when it is named.
Example: a two-week phone audit
Example (fictional): a dental and GP partnership ran 32 mystery calls across two weeks. Average scores were good mid-morning but dropped sharply between 8am and 8.30am, when the queue message said callers were 'being held' without giving a position. The phone supplier switched on the position message and call-back option the system already had. A repeat audit a month later showed the early-morning score rising to match the rest of the day.
NHS England trials found that moving practices to high-quality digital telephony raised patients' ease-of-getting-through scores by 30 percentage points[2]. Not every practice will see a jump like that, but audits often reveal features that are paid for and switched off.
Acting on the findings
- Share overall results with the whole team first, thanking them and focusing on systems.
- Pick the three lowest-scoring areas and agree one change each, with an owner.
- Repeat a smaller audit in six to eight weeks to see whether the changes worked.
- Tell your patient group what you found and what you changed.
- Fold the measures into routine monitoring, alongside call data and complaints.
Audit findings feed naturally into patient journey mapping and into the commitments of a patient charter.
Finally, keep the audit pack: scenarios, scorecard, schedule and results. Running the same scenarios every six or twelve months turns a one-off exercise into a trend line, and a trend line is far more persuasive to partners, commissioners and inspectors than a single snapshot.
Frequently asked questions
Is it ethical to mystery shop a GP practice?
Yes, if the scenarios are admin-only, no real appointments are taken from patients, staff know an audit period is planned and results are used to improve systems rather than to discipline individuals.
Can we record mystery calls?
Recordings identify staff, so you need a lawful basis, a staff privacy notice that covers it and a short retention period. NHS practices cannot rely on legitimate interests for processing in the performance of their public tasks[4], so check the basis with your data protection officer.
How many mystery contacts do we need?
For a small practice, twenty to forty contacts spread across days and times usually shows clear patterns. Repeat a smaller round after making changes.
Who should act as the mystery patient?
Patient group members, staff from a partner practice or an external provider. Avoid anyone your team would recognise.
References
Ipsos (GP Patient Survey provider for NHS England) (2026). 2026 GP Patient Survey results released (opens in a new tab) ↑
NHS England and DHSC (2023). Delivery plan for recovering access to primary care (opens in a new tab) ↑
Independent Healthcare Providers Network (IHPN) (2026). Support for using private healthcare now found across the political spectrum, new report shows (opens in a new tab) ↑
legislation.gov.uk (The National Archives) (2026). UK GDPR Article 6: Lawfulness of processing (opens in a new tab) ↑