Patient Self-Scheduling: Which Visits to Open Online

Let patients book themselves into visits that have a fixed length and need no one to check anything first. Anything that needs a clinical or insurance check goes through a request your staff confirms, or stays on the phone.

Quick answer

1. Sort your visit types into three groups: instant booking, request, phone only.
2. Open instant booking first for established-patient follow-ups and routine visits with a standard length.
3. Take new patients, procedures and long visits as requests until staff approve nearly all of them unchanged.
4. Keep urgent symptoms off the calendar and put an emergency notice above it.
5. Judge the policy after 90 days with four numbers: share booked online, no-show rate by who booked, wrong-type bookings staff caught, and new patients from ads who booked themselves.

How do you decide which visit gets which door?

Ask three things about each visit type. Does it need a human to check something before it is safe or billable? Does it always take the same amount of time? And what does a wrong booking cost: a wasted slot, an insurance problem or a patient who should have been seen sooner?

A yes to the first, a no to the second or a serious answer to the third keeps the visit off instant booking.

An instant booking puts a confirmed time on your schedule the moment the patient clicks. A request collects the patient's preferred times and details, and staff confirm or change them. Phone only means the website shows the number and nothing else. This is a policy about visit types, not about software: what to look for in the tool and where the booking button belongs are separate decisions.

What should each kind of visit get?

This table is my starting position, not a rule. Your specialty, payers and staffing move rows from one column to another.

The telehealth row depends on your state (in Canada, provincial) rules and payer contracts, so confirm them with whoever handles compliance. The row I would change last is the urgent one.

What goes wrong, and what should the scheduler do about it?

Four things go wrong, and a good scheduler has a control for each.

A patient books the wrong visit type. The fix is visit-type rules by patient status, plus a required reason for the visit that staff can read before the day. SigmaMD's help page, as one example, lists an availability setting with "All visits," "First visit only" and "After first visit only," and a reason-for-visit field that can be made required (SigmaMD, checked 2026-10-12). It does not mention a request or approval mode, so I would ask any vendor about that separately.

A patient books a plan you do not accept. Ask for the insurance plan before showing any time, and let the answer decide whether the patient sees a calendar or a request form.

A patient with urgent symptoms books days out. Put wording above the calendar. A patient having an emergency must call 911, and the page should say so before any appointment times appear. For symptoms that are worrying but not an emergency, the same notice should point to your phone line.

Slots are booked and not used. Here you have a minimum notice before the visit, a cap on how far out patients can book, a limit on open future bookings per patient, and appointment reminders. SigmaMD lists an earliest and a maximum time period for booking (SigmaMD, checked 2026-10-12). Deposits are a further control; whether your payer contracts allow them is a question for your billing lead, not the scheduler vendor.

What does published research say?

Less than vendors imply. The study I found is from Mayo Clinic: "Impact of Web-Based Self-Scheduling on Finalization of Well-Child Appointments in a Primary Care Setting: Retrospective Comparison Study" in JMIR Medical Informatics, 2021 (JMIR Medical Informatics via PubMed Central, checked 2026-10-12). It followed 13 months of well-child visits for children aged 2 to 12 at Mayo Clinic sites in Minnesota and Wisconsin.

Its no-show numbers cover finalized appointments, the ones still on the calendar on the day of the visit: 3.07% (28 of 912) for self-scheduled ones and 4.12% (693 of 16,828) for staff-scheduled ones, a difference the authors found not significant. Of the self-scheduling actions, 29.5% happened outside regular business hours, which the study set at 7 a.m. to 5 p.m. on weekdays. The authors note that other studies linked self-scheduling with fewer no-shows, but read their own result as a sign that self-scheduling by itself may not reduce them.

The authors name the limits themselves: one visit type, a mostly White community, only families registered on the patient portal, and a scheduling platform specific to Mayo Clinic. Two details from the methods matter as well: the software did not allow lead times (the gap between booking and visit) over 12 weeks, and patients got text or phone reminders however the visit was booked.

So the study supports a narrow claim: for one routine visit type with no symptoms to assess, self-booking did not raise no-shows there. It says nothing about new patients, procedures or a small practice.

Who handles the requests, and how fast?

A request door is only as good as the reply. Name one person and one backup, pick a reply time you can keep, and say it on the request form. A patient who sends a request on Friday night and hears nothing until Tuesday has been worse served than by a phone call.

Decide what an out-of-hours request gets: an automatic message that says the request arrived and when to expect an answer. Do not let the message promise a time slot. Staff should see each request with the reason, the plan and the patient's status in one place, so approval is a quick read and not a callback.

Treat approval as a measurable step. If staff approve nearly all requests for a visit type without changing them, that visit type has earned instant booking. If they change many, the intake questions are missing something.

How do you judge it after 90 days?

Compute four numbers for the quarter, every one from your own schedule:

• Share booked online = visits booked online ÷ all visits booked
• No-show rate by source = no-shows ÷ visits, for self-booked and staff-booked separately
• Wrong-type rate = bookings staff had to change ÷ self-booked visits
• Ad patients who booked themselves = new patients from ads who booked online ÷ all new patients from ads

Illustrative: the practice in this quarter is imaginary (family medicine, two providers), and every figure is made up rather than taken from a client.

Self-booked no-shows are lower here, but the groups differ: patients who book themselves may be more reliable anyway. A 5% wrong-type rate suggests the intake questions need work before another visit type opens. Ad patients booked themselves at 43.75%, close to the 40% overall share; a figure far below it would suggest the booking button sits where ad visitors do not see it. For how no-shows reach your ad account, see no-shows and conversion adjustments.

What should you ask a scheduling vendor?

• Can I limit visit types by patient status, new or established?
• Can the scheduler ask my intake questions, including insurance, before it shows a time?
• Does it offer a request mode where staff approve, and what does staff see?
• Can I cap how far ahead and how soon a patient can book?
• Can it show an emergency notice above the calendar that I write?
• Can I export bookings with who made them, so I can split no-shows by source?