Monday morning at an orthopedic practice, the main line lights up. A woman six weeks out from a rotator cuff repair wants to know if the ache at night is normal. A man needs to move Thursday's therapy session. A billing question. A request for MRI results. Then, somewhere around the eleventh call, a 58-year-old who has been limping for a year says the words every practice is paying for: "I'd like to see someone about my knee."
Google Ads hears all eleven calls the same way, unless you teach it otherwise. And whatever it hears, it goes out and finds more of.
Google Learns From Whatever You Tell It Counts, Including the Wrong Calls
A conversion in Google Ads is two things at once. It is your scoreboard, the number that says whether the ads are working. It is also the food the bidding system eats. Every time you tell Google "that one counted," it studies the click that led there: the search, the time of day, the device, the neighborhood. Then it bids harder for more clicks that look like it.
Think of the algorithm as a new front-desk hire who learns only from what you praise. Praise every phone call and they conclude that phone calls are the job. They will get very good at producing phone calls. Whether any of those callers is a new patient with a torn meniscus is not their problem, because you never said it was.
This is why tracking comes first in this guide, before budgets, keywords or a single ad. A practice with sloppy tracking doesn't just get bad reports. It gets an account that is actively trained to find the wrong people, and the more it spends, the better it gets at the wrong thing. If your account already counts page views, "any click on a phone number" or every call regardless of length as conversions, that data is misinforming the system right now. The fix is to stop counting it, and sometimes to start the counting over clean.
Eleven of Twelve Practices Send New and Returning Patients to the Same Phone Line
To see how this plays out, we pulled twelve orthopedic practices with live Google Search ads in one example market, Nashville. The set covers big multi-office groups, an academic medical center, hospital service lines, a joint replacement specialty practice, a spine practice, a solo hand surgeon, an orthopedic urgent care chain and a regenerative medicine clinic. Your market will look different. The phone problem won't.
Eleven of the twelve route new and returning patients through the same numbers. The patterns vary. One practice lists a single phone number for all five of its offices. One of the largest groups runs a toll-free central line plus separate lines for its hip program, physical therapy, hand therapy and MRI scheduling, all shown on the same location page. Several show one central number and a string of per-clinic lines. At least seven of the twelve run physical therapy in-house, so therapy scheduling is part of the daily call volume at most of them.
Only one practice in the sample splits the line by patient type. Southern Sports Medicine Institute publishes a new-patient number for each of its two offices and a separate number for existing patients. That is the whole trick, and it costs almost nothing. When the number on your ad's landing page is a number only new patients see, a call to it means something.
Google Can Count a Call. It Can't Tell You Who Was Calling.
Google's own call tracking is the default starting point, and for most practices it is enough. Google recognizes five kinds of call conversion. Someone taps call right on your ad. Someone clicks the ad, lands on your site and dials the forwarding number Google displays there. Someone taps your number on the mobile site. Someone clicks a call button in the ad, which Google counts by estimate. Or you import calls from another system. The first two run on Google forwarding numbers, which swap in on their own so each call ties back to its click.
For both of those, you choose a minimum call length, and only calls at least that long count. We use 30 seconds. Thirty seconds weeds out wrong numbers and hang-ups but keeps voicemails, and a patient who leaves a message at 9 p.m. is still a patient.
Here is the catch. Length tells you a call was real. It doesn't tell you why the person called. A three-minute call about a billing statement and a three-minute call from someone booking a knee consult look identical to Google. No threshold fixes that.
What works is changing the setup, and one practice above already did it: new patients get a number of their own, and that number lives on the pages your ads point to. Your procedure pages, the new-patient page and any landing page built for ads show the new-patient number, with a Google forwarding number sitting on top of it. Your contact page, portal instructions and appointment reminders keep the main line. Existing patients mostly call the number they already have, and the calls Google counts start to mean what you want them to mean.
You'll also hear about third-party call tracking, mainly CallRail and CallTrackingMetrics. They're worth knowing about. Both can record calls, show the exact search that led to each one and track forms alongside calls, which helps a practice that wants to review call quality. Both will sign a business associate agreement, but only on their healthcare plans, so ask before any patient call runs through them. You don't need either one to follow this guide. Four of our twelve example sites showed some kind of call tracking in their page code, including CallRail, a Google forwarding number and WhatConverts. Tools loaded through Google Tag Manager don't show up in that kind of check, so the real number may be higher.
Half the Online Schedulers Book the Appointment on Someone Else's Website
Six of the twelve example practices offer online self-scheduling, and all six send the patient to another company's website to finish it: two to Radix Health, two to Epic MyChart, one to YosiCare and one to Valerie Health. The patient taps "Book Online" on your site and completes the booking somewhere Google Ads can't see by default.
That leaves most practices counting the only thing they can see, which is the click on the button. That's a mistake. A click on "Book Online" proves curiosity. Plenty of people open a scheduler, see the first open slot is three weeks out, and close the tab. Count the button and you train Google to find browsers.
What you want counted is the completed appointment, and the scheduling vendor decides how you get it. It's usually one of these:
- After the booking, the vendor returns people to a thank-you page you host, and your conversion tag fires there.
- The scheduling company lets your tag run on its own confirmation step.
Which one is possible depends on the vendor, so ask them directly. Either way, set up cross-domain measurement in Google Analytics 4, which keeps one visitor counted as one visitor as they move from your site to the scheduler and back, instead of splitting them into two strangers.
The Surgery Happens Months After the Click. The Booked Consult Is the Conversion.
Orthopedic surgery is rarely decided in one visit. A knee patient might see a surgeon, try injections and physical therapy, come back for imaging and schedule surgery months later. In two peer-reviewed studies of spine clinics, 15% to about 42% of new patients ended up in surgery within the study windows, and the gap between them came down to how carefully the referrals were screened.
That timeline matters because Google only counts a conversion within a set window after the click. The default is 30 days, and depending on the conversion type it can go up to 90. A surgery booked four months after the first click falls outside any window Google offers. The booked consult doesn't. It happens days or weeks after the search, it's the step the ad actually caused, and it's the one Google can learn from.
So the new-patient appointment is your conversion. Set the window for forms and bookings to 90 days, and give calls the longest window their source allows. Set counting to "One conversion" per click. Google recommends it for leads, and it's right: a patient who calls twice about the same knee is one patient, not two.
A booked consult is still only a lead. The truest signal is the patient who actually showed up, and that can reach Google too, without a single health detail. Every ad click carries a Google click ID. If your appointment form or scheduler saves it with the request, your staff can send it back after the visit, tagged with a bland label such as "New patient seen" and what the visit was worth. The ID is all Google needs to tie it to the ad; no name, number or diagnosis goes with it. Uploads only count within the conversion window, up to 90 days after the click, which means the attended consult almost always makes it and a surgery booked in the same stretch can go back too. Once those uploads run reliably, they become the conversion Google should optimize for, and calls and forms become the early signals.
A Form That Asks Which Joint Hurts Is Carrying Health Information
Most appointment request forms in this industry ask for the reason for the visit. It's useful for the front desk, and it's a problem for tracking. After a 2024 court ruling threw out part of its guidance, the federal health privacy office still says tracking tools on public scheduling pages that people use without logging in can have access to protected health information in some cases. A form where someone picks "knee" or types "torn ACL" is exactly that kind of page.
Keep the conversion clean:
- Fire it on a plain thank-you page whose web address and event name carry nothing about a body part, condition or doctor. "Appointment request received," not "/thank-you-knee-replacement."
- Don't build remarketing lists from condition or procedure pages. Google limits health audiences anyway, and later chapters show how far.
- Leave enhanced conversions off. Our standard setup turns them on, but Google's customer data policy keeps conversions in sensitive categories out of enhanced conversions, and its example for health is paying for medical services. An orthopedic consult is a medical service, so this is one setting where orthopedic practices skip a step other businesses take.
Have your privacy officer or counsel look over the final setup. This chapter explains how the pieces fit together; it isn't legal advice.
Only Primary Conversions Train Google's Bids. Portal Clicks and Directions Are Decoration.
Google splits conversion actions into two kinds. Primary actions show up in the Conversions column and drive bidding. Secondary actions land in All Conversions, where you can see them and the bidding can't. That split is how you keep the noise out.
In an orthopedic account, these belong in the primary slot:
- Calls of 30 seconds or longer to the new-patient number
- Confirmed online bookings
- Appointment request forms, counted on the thank-you page
- Seen patients uploaded by click ID, once that's running (they become the main goal)
Secondary, for watching only:
- Clicks to the patient portal and bill pay
- Directions clicks
- New-patient paperwork downloads
- Clicks on an email address
One exception: an account too new to have many bookings can promote a lighter signal, such as paperwork downloads, to primary for its first few weeks so Google has something to learn from, then demote it once real consults are coming in.
Then four settings that orthopedic accounts often get wrong:
- Attribution: data-driven, Google's standard model for most conversion actions today. Credit for a booking gets shared among the ad clicks that led to it, rather than all going to whichever came last.
- Enhanced CPC: Google retired it for Search in 2025, but some older conversion settings still show the option. Leave it unchecked; manual bidding in Chapter 7 means manual.
- Installation: put your tags in through Google Tag Manager, so every change lives in one place, and check them with Tag Assistant before any money goes out.
- Cleanup: if an old action counts page views or every call regardless of length, demote it to secondary or remove it. It has been giving the algorithm bad directions for as long as it's been running.
Then test it yourself. Call the new-patient number from your own ad and stay on for 30 seconds. Book a fake appointment through the scheduler and cancel it. Submit the form. If any of those doesn't show up in Google Ads within a day, you found the leak before it cost you a month of budget.
Next: Where the First Orthopedic Dollars Go
With the right calls and bookings counted, Google can finally learn what a new orthopedic patient looks like. Next comes where to show up. Someone typing "knee pain" may be months from surgery; someone typing "orthopedic surgeon near me" is ready to book. Chapter 2 shows which searches get your first dollars, and why.
The guide home lays out all sixteen chapters, from tracking through surgeon brand campaigns, injection ad rules and walk-in injury care.
When an orthopedic group brings in our Google Ads management team, day one looks like this: we dial every number on the site, book a fake consult through the scheduler, and sort out what counts before any bid changes. Doing it yourself? Make those three test conversions this week, before you spend another dollar.
