Run Facebook and Instagram adsfor your clinic

What you will have at the end.

One campaign with a structure that can actually learn, creative built on a hook that survives three lines, and a lead flow where every enquiry reaches a human within minutes.

Search catches people who are already looking. Social advertising reaches people who were not, which makes it the right tool for a service nobody thinks to search for, and for filling the Tuesday afternoon slots that never fill on their own.

It is also the platform where clinic owners most often build something that cannot possibly work: nine ad sets, each with a hundred dollars a month, each trying to learn independently, none of them getting enough data to learn anything.

Structure, and why less is more here

1. One campaign, one ad set, many ads

   This is counterintuitive if you come from search, and it is the single most important structural decision on this platform. The system optimizes at the ad set level. Splitting your budget across many ad sets splits the learning, and a clinic budget is not large enough to survive that.

   Put the budget in one ad set and let the system choose among your ads. Ten or fifteen ads in one ad set is normal and healthy.

   —A second ad set is justified when you genuinely need a different geography, a different language, or a different budget you would never let the system move.

   —It is not justified by wanting to compare two audiences. The system will do that inside one ad set, faster.

   —If you have two service categories with different patients, that is two campaigns, not two ad sets in one campaign.

1. Set the geography tightly and check the delivery

   Location is the one targeting control that still behaves. Set a radius around the clinic matching where patients actually come from, and choose the option for people living in the area rather than people recently in it.

   After a week, pull the delivery breakdown by location, age and placement. Compare it to what you set. The gap is the thing to know.

2. Pick the objective that matches where the enquiry lands

   If you want the person to book on your site, optimize for the conversion event on your site. If you want a phone call back, a native lead form on the platform will fill faster and with lower-quality enquiries.

   Both are legitimate. What is not legitimate is optimizing for clicks or reach and then complaining that nobody booked.

   —Conversion on site: fewer, better, requires working tracking and a fast site.

   —Native lead form: more volume, lower intent, requires a human calling back within minutes rather than hours.

   —Engagement: only for building an audience you will advertise to later, never as the main play.

The creative is the targeting now

1. Write the first three lines as if they are the whole ad

   On a phone, roughly three lines show before the more link. Almost nobody taps it. Everything that qualifies the reader and states the promise has to live there.

   Name the person, name the problem, name what you are offering. In that order, in plain language.

   —Good: If you have been putting off a dental visit for years, we have a clinic built for exactly that.

   —Weak: At our clinic, we believe every patient deserves compassionate care.

   —Never open with your clinic name. They do not know it yet and it buys you nothing.

2. Make the image say the thing in four seconds

   Someone scrolling gives an ad about four seconds. The image has one job: make the promise legible before they have decided anything. A headline on the image, short and literal, outperforms a beautiful picture with no words on almost every clinic account we have seen.

   —One short headline on the image, naming the outcome or the service. Not clever, not a riddle.

   —Real photos of your actual clinic and actual team beat stock, when you have them.

   —Avoid anything that could read as a medical claim or a before and after of a body. It will be rejected and repeated rejections affect the account.

   —Test a plain graphic version against a photo version. Which wins varies by clinic and you cannot predict it.

   In practice

   Refresh creative every three to four weeks. Social ad fatigue in a small local audience is fast, and a clinic radius is a small audience.

3. Run enough variants to find the one

   Build the ads as a grid: three or four hooks crossed with two or three images. That is nine to twelve ads, all in the same ad set, and the system finds the combination that works far faster than you can reason your way to it.

   —Change one element at a time across the grid so you can read the result.

   —Give it enough time and budget that the winner is a result and not noise.

   —Kill the clear losers after two weeks, not two days.

Lead forms, and the traps specific to healthcare

1. Set the form up so it works on the first try

   Native lead forms fail for a small set of predictable reasons. Walk them before launching rather than debugging live.

   —The page terms must have been accepted by an administrator of the page itself. This is a separate acceptance from your account terms and it is the most common reason a form will not publish.

   —The privacy policy link is required, and it must actually load. A link to a page that 404s blocks the form.

   —The form language has to be one of the supported locale codes, and the correct one for your region. An invalid code is rejected without a useful error.

   —Ask for the minimum: name, phone, and which service. Every extra question costs completions and none of them qualify anybody.

   —Never request health information, insurance numbers or symptoms in the form.

2. Get the enquiry to a human within minutes

   A native lead form enquiry is warm for a very short time. The person tapped a button while scrolling. If you call them ninety minutes later they have forgotten, and if you call them the next morning they have booked elsewhere.

   Build the notification before you turn the ads on. Not after the first week, when you discover there are forty enquiries sitting in a dashboard nobody opens.

   —Every submission triggers an email or a text to a named person, immediately.

   —Enquiries also land in one list you can see, so nothing depends on an inbox.

   —Somebody calls back the same day, and twice if the first call is not answered.

   —Track how many enquiries get called, not just how many arrive.

   In practice

   Most clinics that conclude social advertising does not work have a callback problem, not an advertising problem. The enquiries arrived and nobody phoned.

3. Make sure a booking counts once

   Double counting is the standard failure on this platform. The pixel fires on the page, the conversion interface fires again from the server, and the platform reports two.

   Check it by booking a test appointment yourself and watching how many conversions appear.

   —One booking must produce one conversion. Not two, not three.

   —If you send events from both the browser and your server, they must share an event identifier so the platform can deduplicate them.

   —Purchases and leads are different event types. Report them separately rather than adding them into one number.

Custom audiences, and where they break

1. Build the two audiences worth having

   Retargeting your own visitors and excluding your existing patients are the two audience moves that reliably earn their keep for a clinic. Everything else is optional.

   —People who visited a service page and did not book, over the last thirty days.

   —Your existing patient list, uploaded and then excluded, so you stop advertising to people who already come to you.

   —Handle the patient list carefully. A list of clinic patients is health information by implication, and uploading it is a decision to make deliberately, with consent that covers it.

   In practice

   Website custom audiences depend on a separate terms acceptance. If that acceptance is missing, every website-based audience is blocked at once, which looks like a technical fault and is actually a checkbox.

When it goes sideways