Key takeaways
- Six measures cover most of what matters: answer rate, abandonment, speed to answer, the long tail of waits, call-back completion and repeat calls.
- Count abandonment carefully: callers who hang up during the greeting are different from those who give up in the queue.
- Look at the busiest hour and the worst day, not just the monthly average.
- Set targets from your own baseline, review weekly, and pair each number with one action.
Why averages mislead
Most digital phone systems produce a monthly report with an average wait and a total number of calls. Those two figures hide most of what patients experience. An average wait of two minutes can mean everyone waited two minutes, or that afternoon callers got through instantly while 8am callers waited twenty minutes and many gave up.
Patients judge the practice on the call they made, usually at the busiest time. In the 2026 GP Patient Survey, 56.9% said it was easy to contact their practice by phone[1], which also means a large minority did not. The measures below are designed to show where that minority comes from, so that reception leads can act on it.
The six measures
| KPI | How to calculate | Where the data comes from | Warning sign |
|---|---|---|---|
| Answer rate | Calls answered by a person or resolved by self-service, divided by calls offered | Phone system call report | Falls on Mondays or after bank holidays |
| Abandonment rate | Calls ended by the caller after joining the queue, divided by calls that joined the queue | Queue report, excluding hang-ups during the greeting | Rises sharply in the first hour |
| Average speed to answer | Total queue time for answered calls, divided by answered calls | Queue report | Rises steadily through the week |
| Long-wait share | Calls that waited longer than your threshold, divided by calls answered | Wait-time distribution report | Any growth, even with a stable average |
| Call-back completion | Call-backs completed within the promised window, divided by call-backs requested | Call-back log | Below the level you promise patients |
| Repeat callers | Numbers that call again within two working days, divided by unique numbers calling | Call detail records by caller number | Rises after a change in process |
Keep definitions fixed once you choose them. If you change what counts as abandoned, the trend breaks and nobody can tell whether things got better.
Getting abandonment right
Abandonment is the most misunderstood measure. Some callers hang up during the greeting because they hear the message they needed, for example that the online form is open or that results are not ready. Others hang up after the menu because they could not find the right option. Others give up after waiting in the queue. These are three different problems with three different fixes.
NHS 111 reports the share of calls abandoned after its call-steering menu separately: 4.6% in July 2026[2]. Do the same. Split your abandonment into during greeting, after menu and in queue, and track each one.
- High abandonment during the greeting: check whether the greeting is too long or whether callers are getting the answer they need. Both are possible.
- High abandonment after the menu: the options are unclear or too many. See our guide to IVR menu design.
- High abandonment in the queue: not enough people answering at peak, no position message, or no call-back option.
Speed to answer and the long tail
NHS 111, a service designed around call handling, had an average time to answer of 82 seconds in July 2026, with 75.6% of calls answered within 60 seconds[3]. Practices are not call centres and will not match that at 8am, but the way NHS 111 reports is useful: an average plus the share answered within a threshold.
Pick a threshold that matters to your patients, such as five minutes, and report the share of calls that waited longer. A practice can improve its average while the longest waits get worse, for example by answering quick afternoon calls faster. The long-wait share catches that.
Call-backs and repeat calls
Digital telephony features promoted by NHS England include queue position messages and an option for patients to be called back[4]. A call-back option only helps if the call-backs happen. Measure completion within the window you promise, and count failed attempts separately from successful ones. A call-back that rings once and gives up is not a completed call-back.
Repeat callers are the best phone-based proxy for whether problems are being resolved. If the same number rings again within two working days, the first call probably did not end with a clear outcome. Our guide to first-contact resolution explains how to reduce repeat calls.
Context measures that explain the KPIs
The six KPIs tell you how well calls are handled. They make sense only next to a few measures of demand and capacity, which explain why a week went well or badly.
- Calls offered per hour and per day, plotted for each weekday, to show where demand peaks.
- Unique callers per day, which strips out repeat dialling and shows true demand.
- Staffed phone hours per hour of the day, from the rota, set against calls offered.
- Calls per thousand registered patients each week, so you can compare across sites or with neighbouring practices in your PCN.
- Online requests received in the same hours, because a rise online often explains a fall on the phone.
When answer rate falls, the first question is whether demand rose or staffed hours fell. These measures answer that in a minute and stop the team from blaming itself for a problem that is really about the rota.
Example: one week's figures
Example: a fictional practice, Brookside Surgery, with about 9,000 patients, pulled one week of data. It received 2,400 calls. Of the 2,150 that joined the queue, 1,870 were answered, 210 were abandoned in the queue and 70 chose a call-back. The abandonment rate in the queue was about one in ten, but on Monday between 8am and 9am it was closer to one in three.
Average speed to answer was four minutes across the week, but 18 minutes in Monday's first hour. Of 70 call-backs requested, 61 were completed within the promised window. Repeat callers within two working days were about one in eight unique numbers. The reception lead chose one action for each problem: two extra people on the phones on Monday from 8am to 9.30am, a check of the call-back list at 11am and 3pm, and a closing script that confirms what happens next.
Dental practices and private clinics
The same six measures apply, with two additions. First, missed calls outside opening hours: count calls to the main number when the practice is closed and how many of those callers were later contacted. Second, enquiry conversion: of new-patient or treatment enquiries by phone, how many led to a booked appointment within a week.
For a private clinic, conversion is often the measure the owner cares about most, but it depends on the others. A clinic that answers quickly and calls back missed enquiries the same day gives its team far more chances to book. Review missed enquiries daily rather than weekly; a call returned the next morning reaches far fewer people than one returned within the hour.
Turning numbers into action
Agree definitions
Write down exactly how each KPI is calculated and which report it comes from. Share it with anyone who reads the figures.
Take a baseline
Collect four weeks of data before setting any target. Note unusual events such as a bank holiday or an outage.
Set targets from the baseline
Aim for steady improvement on your own figures. A target copied from another service that you have no way of reaching demoralises staff.
Review weekly, by hour and day
A ten-minute weekly look at the busiest hour and the worst day is worth more than a long monthly report.
Pair each number with one action
If a KPI slips, agree one change, try it for two weeks and check the effect.
Share results with the team
Show reception staff the trend and what changed. Do not use the figures to rank individuals.
What not to measure
Some measures do more harm than good at practice level. Average call length per receptionist pushes staff to rush vulnerable patients. Calls answered per person ignores the difference between a booking call and a distressed patient. Neither tells you whether patients got what they needed.
The national picture is that most patients now succeed in making contact: ONS figures cited by the government put success at 96.3% for patients who tried to contact their practice in the previous 28 days[5]. The job of practice KPIs is to find the patients and the hours where that is not true, and fix them. Pair the figures with mystery shopping your own phone line to hear what the numbers cannot show.
Frequently asked questions
What is a good abandonment rate for a GP practice?
There is no national target for practices. Measure your own baseline, split by greeting, menu and queue, and aim to cut queue abandonment at your busiest hour first.
Should we count calls answered by an automated option?
Count them separately. A caller who got their answer from a recorded message was helped, but staff answer rate and self-service answers tell you different things.
How often should KPIs be reviewed?
Weekly for the busiest hour and day, monthly for trends. Short, frequent reviews lead to quicker fixes.
References
Ipsos (GP Patient Survey provider for NHS England) (2026). 2026 GP Patient Survey results released (opens in a new tab) ↑
NHS England (2026). Statistical Note: Integrated Urgent Care Aggregate Data Collection (IUC ADC) July 2026 (opens in a new tab) ↑
NHS England (2026). Statistical Note: Integrated Urgent Care Aggregate Data Collection (IUC ADC) July 2026 (opens in a new tab) ↑
NHS England and DHSC (2023). Delivery plan for recovering access to primary care (opens in a new tab) ↑
GOV.UK (DHSC) (2025). Millions more appointments as more than 2,000 extra GPs recruited (opens in a new tab) ↑