Key takeaways
- There is no reliable independent benchmark for how long patients will hold. Your own abandonment-by-wait data shows your real limit.
- Measure three different waits: time in the queue before answer, holds inside a call, and the time to a callback.
- Set targets just below the point where abandonment jumps in your data, and offer a callback before callers reach it.
- Most hold time comes from a few long call types and a few peak hours. Fix those first.
What is known, and what is not
Ask how long patients will stay on hold and you will find many confident answers online, mostly from phone and contact-center vendors measured on their own customers. There is no independent national study of hold tolerance in US medical practices that we could verify. That is not a reason to guess. Your phone system already records how long each caller waited and whether they hung up, which is a better answer for your patients than any average.
What is clear is that access is on practice leaders' minds. When MGMA asked about their top patient access focus for 2026, phone access (22%) sat beside no-shows, online scheduling and wait times, which together accounted for nearly all responses[1]. MGMA also notes that the phone remains the real front door for many patients[2].
Source: Medical Group Management Association (MGMA), 2025[1]
Show the data
| Item | Value |
|---|---|
| No-shows | 27% |
| Online scheduling | 24% |
| Phone access | 22% |
| Wait times | 21% |
Why patients hang up sooner than they used to
Patients now have other places to take a simple question or a minor illness. In 2024, 19.0% of people in the US had at least one retail health clinic visit in the past year[3]. Some turn to AI tools: among adults who use AI for health information, 65% said wanting quick and immediate advice was a major reason[4], and 18% said a major reason was not having a regular provider or not being able to get an appointment[5].
None of these numbers measures hold time directly. They do show that a caller left on hold has alternatives, and that speed is part of what patients expect. A long hold is not only an annoyance; for a new patient or a sick established patient, it can be the moment they choose somewhere else.
Three waits worth measuring
| Measure | What it means | Where to get it |
|---|---|---|
| Queue wait | Time from the end of the phone menu until a person answers | Phone system call detail report |
| Abandonment by wait | Share of callers who hang up, grouped by how long they had waited | Call detail report, grouped into wait bands |
| Hold inside a call | Time a caller is placed on hold after being answered | Phone system hold report, or a sample of recordings |
| Callback time | Time from a voicemail or callback request to the return call | Voicemail timestamps against outbound call log |
| Repeat callers | Same number calling again within the same day | Call detail report, grouped by caller number |
Most practices only look at average speed of answer. Averages hide the problem: a morning with half the calls answered at once and half waiting six minutes can still show a respectable average. The distribution matters more than the mean.
Build your abandonment curve
Export four weeks of call detail
For every inbound call you need the queue wait in seconds and whether it was answered or abandoned. Leave out calls that ended inside the phone menu; they are a menu problem, not a hold problem.
Group calls into wait bands
Use bands such as under 30 seconds, 30 to 60 seconds, 1 to 2 minutes, 2 to 3, 3 to 5, and over 5 minutes.
Work out the abandonment share in each band
Abandoned calls in the band divided by all calls that reached that band.
Find the jump
Look for the band where the share of callers hanging up rises sharply. That is your practical hold limit.
Split by caller type if you can
Callers choosing the new-patient option often give up sooner than established patients. A separate curve for each menu option shows where to protect access first.
Example: one practice's curve
| Queue wait | Example calls reaching this wait | Example share hanging up in this band |
|---|---|---|
| Under 30 seconds | 2,400 | 2% |
| 30 to 60 seconds | 900 | 4% |
| 1 to 2 minutes | 600 | 7% |
| 2 to 3 minutes | 380 | 16% |
| 3 to 5 minutes | 240 | 25% |
| Over 5 minutes | 120 | 38% |
Example (fictional): in this practice, abandonment more than doubles once callers have waited two minutes. The manager sets two targets: answer most calls within one minute, and offer a callback option at about ninety seconds, before callers reach the band where they start to leave. She also finds that almost all waits over three minutes happen on Monday and Tuesday between 8 and 10 a.m.
Setting targets you can keep
A target is only useful if the schedule can meet it. Write targets in the form “X% of calls answered within Y seconds”, using your own curve for Y and your current performance for X, then raise X gradually.
- Speed of answer: set Y just below the band where abandonment jumps in your data.
- Holds inside a call: ask permission, give a time and check back at a fixed interval set by the team.
- Callback offer: trigger it automatically before callers reach your jump point, if your system supports it.
- Callback time: set a same-day standard for messages left before a cut-off time, and next morning for later messages.
- Peak-hour target: track the busiest two hours separately. They decide how patients feel about the practice.
Do not publish an expected wait time to callers unless the system calculates it live. A recorded “your wait is about two minutes” that turns into six does more harm than saying nothing.
Cutting hold time without hiring
Hold time is mostly a mismatch between when calls arrive and when people are free. A few changes target that mismatch directly:
- Move long calls off the main queue. MGMA's poll found eligibility and prior authorization the most time-consuming phone task for staff[6]. A separate line for insurance and authorization questions stops one long call from blocking five short ones.
- Staff the peak, not the average. Staggered start times and phones-only cover in the first two hours do more than an extra person in the afternoon.
- Offer callbacks. A callback request turns waiting time into scheduled work at a quieter hour.
- Remove avoidable calls. Status checks for results and refills inflate the queue at every hour.
- Shorten the phone menu. Every option before the queue adds seconds that the caller counts as waiting.
- Use overflow cover. When the queue passes your limit, send calls to a back-office phone, an answering service or an SI Employee such as Ava, which can answer general questions and pass anything patient-specific to the staff line you choose.
The hold nobody measures: inside the call
Queue reports miss a second kind of wait: the hold after someone answers, while staff check the schedule, look for a nurse or ask billing a question. Many systems report it only as part of total talk time. Listen to a sample of recorded calls from your busiest morning and note each hold: why it happened, how long it lasted and whether the caller was asked first.
- Holds to find information usually mean a screen or a list is missing at the desk, such as the insurance plans accepted or the providers taking new patients.
- Holds to reach a colleague usually mean a routing problem: the call should have gone to that person in the first place.
- Holds to get approval, for example to double-book, mean the desk needs a written rule it can apply without asking.
- Repeated holds in one call are a sign the call should become a callback with a set time.
Fix the cause of each common hold and both waits shrink: the caller on the line waits less, and the callers in the queue get answered sooner because the line frees up faster.
Callers who wait longer through no fault of their own
Some callers face extra steps before the conversation can start: adding an interpreter, a relay operator introducing the call, or a caregiver conferencing in a family member. These calls take longer by nature, and a strict hold target can push staff to rush them. Track them separately, keep the interpreter number on the desk so it can be added in seconds, and make sure nobody treats the pause at the start of a relay call as a dead line.
What callers hear while they wait
The wait feels shorter when it is useful. A rotating set of short messages can answer common questions, mention the portal for refills and remind callers how to reach emergency help. Avoid repeating “your call is important to us” every twenty seconds; it tells callers nothing and makes the wait feel longer. Our guide to on-hold messages has scripts, and our phone tree guide covers how to keep the menu short.
Review the curve monthly. When you change staffing or the menu, re-run the same four-week export so the comparison is fair. For the calls that create most of the queue, start with the top reasons patients call and our plan to reduce call volume without hurting access.
Frequently asked questions
How long will patients stay on hold?
It varies by practice and caller, and there is no verified national benchmark. Group your own calls by how long they waited and find the band where hang-ups jump; that is your practical limit.
What is a good hold time for a doctor's office?
One you can meet at your busiest hour that sits below your own abandonment jump point. Write it as a share of calls answered within a set number of seconds, and track the peak hours separately.
Should we offer callbacks instead of hold?
Yes, if you can return calls the same day. Offer it before callers reach the wait where they usually hang up, and make sure someone owns the callback list.
Does on-hold music reduce hang-ups?
Music alone makes silence less confusing, but short useful messages, a callback offer and shorter waits make the bigger difference.
References
Medical Group Management Association (MGMA) (2025). Patient access priorities for 2026: Tackling wait times, phones, no-shows and more (opens in a new tab) ↑
Medical Group Management Association (MGMA) (2026). Phones are still a bottleneck costing medical practices time they can't afford (opens in a new tab) ↑
CDC National Center for Health Statistics (2024). Urgent Care Center and Retail Health Clinic Use: United States, 2024 (NCHS Data Brief No. 562) (opens in a new tab) ↑
KFF (2026). KFF Tracking Poll on Health Information and Trust: Use of AI For Health Information and Advice (fielded Feb 24 - Mar 2, 2026) (opens in a new tab) ↑
KFF (2026). KFF Tracking Poll on Health Information and Trust: Use of AI For Health Information and Advice (fielded Feb 24 - Mar 2, 2026) (opens in a new tab) ↑
Medical Group Management Association (MGMA) (2026). Phones are still a bottleneck costing medical practices time they can't afford (opens in a new tab) ↑