Automate the task that happens most often, needs the least judgment, and doesn't require patient information to leave the systems you already run. For most small practices that's appointment reminders and confirmations, and surprisingly often the first version costs $0, because your EHR already has some reminder system built in or offers a plugin that adds one. Custom automation—the kind that recovers a cancelled slot, chases an overdue recall, or gets a referral booked this week instead of next month—runs $2,500–$5,000 in our pricing, and it earns that only after the free layer is switched on.
Here's how to rank the rest of the front desk, and what each rung typically costs.
How do you decide what to automate first?
Three questions, asked about every repetitive task in the office:
- How often does it happen? Something done dozens of times a week is worth automating even if each instance is small. Something done twice a month is not, however painful it is.
- Does it need judgment? If a competent new hire could do it from a one-page checklist on their first day, it's a candidate. If the answer depends on reading a person, it isn't—yet.
- Does patient information have to leave your systems? A task that stays inside your EHR is simple. A task that sends patient information to a new vendor needs a Business Associate Agreement and a conversation about where the data lands. That's not a reason to skip it, just a reason to sequence it later.
Score honestly and the order mostly writes itself. The arithmetic that makes it real is minutes per instance times instances per week. A front desk that spends ten minutes on each of thirty reminder calls has spent five hours—most of a workday—on something a machine does better and never forgets.
What to automate first: the usual order
| Task | Usually built into the EHR? | Patient info leaves your systems? | Typical first move |
|---|---|---|---|
| Reminders and confirmations | Often, or via a plugin | No | Switch it on |
| Recalls and waitlist backfill | Rarely done well | Only if a new vendor is involved | Custom build |
| Referral intake and slot matching | Rarely | Yes | Custom build, early |
| Intake forms before the visit | Often | No, if it's the EHR's own tool | Switch it on |
| The same ten phone questions | No | No, for general questions | General-questions chatbot |
| Other inbound faxes and documents | Partially | Yes | Later, deliberately |
Reminders and confirmations. Check your EHR first: often it has some reminder system built in, or a plugin marketplace that adds one. Turn that on before buying anything. Our reminders guide covers what a message may say and the consent it needs. A practice paying for custom reminders it could have switched on for free bought the wrong thing.
Recalls and waitlist backfill. This is where custom work earns its price. Built-ins remind; they rarely recover. The patient who never rebooked their annual, the cleaning six months overdue, the 2 PM cancellation that could have been filled from a waitlist—that's revenue your current tools leave on the table, and it's mechanical enough to automate well. It matters most where the schedule swings: a practice near Lake George sees a very different July than January, and a waitlist that backfills itself is worth the most in exactly those weeks.
Referral intake and slot matching. The same logic, one step upstream. Anything that pulls revenue forward is low-hanging fruit, and referrals are the clearest case: one that arrives by fax and sits in a tray for three days is a patient booked next month instead of this week. Reading referrals as they arrive and matching them against open slots on the schedule is mechanical enough to automate well, and in our experience it pays back about as fast as recall work. It does mean patient information flowing through new software, so the vendor paperwork comes first—but that's a reason to do the paperwork early, not a reason to wait.
Intake before the visit. If patients still fill out a clipboard in the waiting room and someone retypes it, that's the most visible waste in the building. Many EHRs offer some form of digital intake now, built in or as a plugin; use theirs before buying another vendor's, because it keeps patient information inside a system you already have an agreement with.
The same ten phone questions. Hours, parking, whether you take a plan, whether you're accepting new patients. A bot that answers only general questions never touches patient information, which keeps it cheap and low-risk—the HIPAA chatbot guide draws that line in detail. Count a week of voicemails first: if they're mostly reschedules, you want the reminders work above, not a chatbot.
Other faxes and documents. Records requests, prior authorizations, lab reports—still how a lot of paper arrives, and genuinely automatable. Unlike referrals, though, most of it doesn't pull revenue forward, so it belongs after the easy wins, with the vendor paperwork done properly.
What shouldn't be automated yet?
Anything clinical—symptoms, results, medication questions—routes to a person, quickly and obviously, with no attempt to answer first. That's the same boundary we put around reminders and chatbots, and it isn't negotiable.
Two less obvious ones. Collections and billing disputes: the words matter, the tone matters, and one wrong automated message costs you a patient. And anything fed by messy data—if the schedule has three different ways of marking a no-show, clean that up first, or the automation will faithfully do the wrong thing at scale.
What does front-desk automation cost?
Surprisingly often, the honest starting price is $0, for the reminder and intake tools already in the subscription you're paying for. Do those first; an afternoon in the settings is the best-returning project on this list.
Custom automation in our pricing is an automation sprint at $2,500–$5,000—a fixed-scope, two-week build of the highest-payback items, which for a practice is usually the recall, waitlist, and referral work. A general-questions chatbot is a separate build at $3,000–$10,000. Running costs for either are typically $20–$150 a month, disclosed up front. Any vendor that touches patient information signs a Business Associate Agreement before anything goes live.
I'm not a lawyer, and this isn't legal advice. HIPAA carries real penalties, so make sure your counsel is part of any project that moves patient information, and that they're satisfied with the plan before it goes live.
How do you know it worked?
Count before you start: empty slots per week, recalls overdue, voicemails per day, hours the front desk spends on the phone. Count again a month in. If the numbers didn't move, stop—your money, our honesty. If they did, automate the next rung.
That sorting is most of what we do in automation work for practices: find the repetitive thing that's actually costing money, automate exactly that, and keep patient information handled the way the law expects. Our AI and automation service starts with a free half-hour walk through your front desk, and "turn on what you already pay for" is a real possible answer.