Key takeaways
- Test calls show what patients actually hear, which call statistics cannot.
- Tell staff in advance that test calls will happen, never use real patient details, and score the system as well as the person.
- Use five or six realistic scenarios, including an urgent one, and call at different times of day.
- Share results as a team, lead with what went well and fix process problems before coaching individuals.
Why test your own line
Public satisfaction with general practice and NHS dentistry is low. In the 2025 British Social Attitudes survey, 35% of people were satisfied with GP services and 45% dissatisfied; for NHS dentistry the figures were 22% and 54%[1]. Much of that judgement is formed on the phone, before a patient ever meets a clinician.
- Satisfied
- Dissatisfied
Source: Nuffield Trust and The King's Fund, 2025[1]
Show the data
| Item | Satisfied | Dissatisfied |
|---|---|---|
| GP services | 35% | 45% |
| NHS dentistry | 22% | 54% |
NHS England has noted that long waits on the phone affect callers' interactions with reception staff[2]. By the time a frustrated patient reaches a person, the conversation is already harder. Call statistics tell you how long people waited. A test call tells you what they heard while waiting, what the receptionist said, and whether the patient would have known what to do next.
Ground rules before you start
- Tell the whole team that test calls will take place over the coming month, why, and how results will be used. Surprise testing damages trust and is not needed to get useful results.
- Never use a real patient's name, date of birth or NHS number. Use an agreed fictional identity that reception can recognise once the call ends, or end the call before identity checks.
- Do not book real appointments. End the call politely before a slot is confirmed, or cancel immediately.
- Score the whole experience, including the greeting, menu and queue, not just the receptionist.
- Keep results at team level. The purpose is to improve the service, not to find someone to blame.
- Avoid testing during known crisis points, such as a phone outage or severe staff shortage, unless that is what you want to understand.
Team message: 'Over the next four weeks, a few test calls will come in so we can hear our phone line the way patients do. They will not use real patient details. We will share what we learn together and fix the system issues first.'
Choosing the scenarios
Pick scenarios that reflect your real calls and the moments that matter most. Five or six is enough for one round. Write a short brief for each so different callers behave in the same way.
- Routine appointment request from a patient who cannot use the online form.
- A parent calling about a feverish child, to test urgent signposting and same-day handling.
- A question about a repeat prescription that has not arrived at the pharmacy.
- A new patient asking how to register, or for a dental practice, whether NHS places are available.
- For dental practices: a caller with toothache who has no regular dentist. Many such patients find a practice through an online search[3].
- For private clinics: an enquiry about a treatment, its suitability and how to book a consultation.
Include one call made at 8am on a Monday, one at lunchtime and one late in the afternoon. The same team can sound very different at different times.
The scoring sheet
| Stage | What to check | Score |
|---|---|---|
| Getting through | Rings answered or queue joined; no engaged tone; position or wait message | 0 to 2 |
| Greeting and menu | Practice name clear; urgent route early; options easy to follow; under about a minute | 0 to 2 |
| Wait | Time in queue recorded; call-back offered if long | 0 to 2 |
| Opening | Receptionist gives name or practice; friendly tone; listens before asking questions | 0 to 2 |
| Understanding the need | Asks the right questions for the request; no unnecessary detail sought | 0 to 2 |
| Safety | Spots the urgent scenario; gives correct urgent wording; does not delay | 0 to 2 |
| Outcome | Clear next step, time and who will act; offers choice where possible | 0 to 2 |
| Close | Checks understanding; asks if anything else is needed | 0 to 2 |
Add a free-text box for 'What would a patient remember about this call?' That one sentence is often the most useful part of the sheet. Also record the time of call, the wait in seconds and any system problems such as dropped calls.
Running the round
Recruit callers
Use people the team does not know well: a PCN colleague, a patient participation group member who agrees to the rules, or a manager from a partner practice. Swap with a neighbouring practice if you can.
Brief them
Give each caller the scenario, the fictional identity, the time to call and the scoring sheet. Ask them to fill it in straight after the call.
Spread the calls
Aim for two or three calls per scenario across different days and times, giving roughly fifteen calls in total.
Collect and anonymise
Put all sheets into one summary. Remove names of staff from the summary; you are looking for patterns.
Compare with your data
Check the call statistics for the same days. If test callers waited far longer than the average suggests, your average is hiding a peak.
Problems test calls usually uncover
Across practices, the same handful of issues come up again and again. Knowing them in advance helps callers notice them and helps you plan fixes.
- Out-of-date recorded messages: last winter's flu clinic, a closed branch site, an old opening time.
- Urgent signposting buried after a long list of instructions about online forms.
- Menu options with internal names, such as 'secretaries' or 'admin', that patients do not understand.
- Call-back options that drop the request or call back from a withheld number that patients do not answer.
- Calls transferred to a line that rings out, especially after lunch or near closing time.
- Receptionists asking for clinical detail they do not need, or not asking enough to route the call safely.
Adapting the sheet for private clinics
For a private clinic, the phone call is often the first sales conversation, so add two stages to the scoring sheet. The first is information: did the caller get a clear explanation of the consultation process, what happens next and how fees are explained, without pressure? The second is follow-through: if the caller did not book, was their contact taken with permission and were they followed up as promised?
Keep the clinical boundary in view. A receptionist should never suggest a treatment will suit someone before an assessment, and good practice is to say so plainly. Score that as a safety item, in the same way as urgent signposting in a GP practice.
Sharing results kindly
How you share results decides whether the next round is welcomed or dreaded. Start with what went well, quoting specific good moments. Then present system problems: a greeting that takes ninety seconds before the urgent route, a menu option that sends prescription calls to the wrong queue, a call-back that never came. Fix these first, because they are not anyone's fault and they affect every call.
Only then look at conversation skills, as a team. 'Several calls ended without a clear time for the call-back' is a training point for everyone, not a criticism of one person. If an individual needs support, discuss it privately and use it as coaching, not discipline. Our guide to using call recordings for coaching covers fair feedback in more detail.
Example: what one round found
Example: a fictional practice, Meadowbank Surgery, ran fifteen test calls over three weeks. Receptionists scored highly on tone and on spotting the feverish child. The weakest stages were the greeting, which took over a minute before mentioning urgent help, and the outcome, where four calls ended with 'someone will ring you' and no time.
The practice shortened the greeting, moved the urgent wording to the first fifteen seconds and introduced a closing script that gives a time window for every call-back. A second round two months later showed the outcome score rising across all scenarios. Patients' own view of phone access nationally has been improving[4]; testing showed the practice where its own line stood against that trend.
Checklist
- Team told in advance, with the purpose and how results will be used.
- Fictional identities agreed; no real patient data used.
- Five or six scenarios written, including one urgent call.
- Scoring sheet prepared and callers briefed.
- Calls spread across days and times, including 8am Monday.
- Results anonymised and compared with call statistics.
- System fixes made first, then team coaching points agreed.
- Next round scheduled.
Repeat the exercise twice a year, and after any big change such as a new phone system or menu. Combine it with the call handling KPIs and the work on first-contact resolution, so you can see whether the numbers and the experience move together.
Frequently asked questions
Is it fair to mystery shop our own reception team?
Yes, if staff are told in advance, real patient data is never used and results are shared at team level to improve the system rather than to blame individuals.
How many test calls do we need?
About fifteen calls across five or six scenarios and different times of day is enough to show patterns.
Should we record the test calls?
If your system records all calls and staff know this, the test calls will be recorded too. The scoring sheet is usually enough; use recordings only for agreed coaching.
References
Nuffield Trust and The King's Fund (2025). Public satisfaction with the NHS and social care in 2025: results from the British Social Attitudes survey (opens in a new tab) ↑
NHS England and DHSC (2023). Delivery plan for recovering access to primary care (opens in a new tab) ↑
Office for National Statistics (2026). Experiences of NHS healthcare services in England: September 2026 (opens in a new tab) ↑
Ipsos (GP Patient Survey provider for NHS England) (2026). 2026 GP Patient Survey results released (opens in a new tab) ↑