Google Ads for Doctors and Clinics: Advertising Healthcare Without Overpromising

The short answer

In most verticals, advertising rules are friction you route around on the way to a strategy. In healthcare they are the strategy. What a clinician is permitted to claim, what may be tracked about a visitor, which audiences may be built, and what counts as an honest representation of a treatment are not compliance chores bolted onto a campaign — they determine the shape of the campaign itself. A post that treats them as annoyances is worse than useless to someone whose license is attached to the account.

So the honest version is this. A clinic account is narrower than a comparable account in any other local service category. It bids on fewer terms, over a smaller radius, with plainer copy, and it measures something further downstream than most advertisers ever bother to measure. It gives up the two levers that other advertisers rely on most — aggressive claims and behavioral remarketing — and it has to make up the difference with proximity, reputation and speed of response. That is not a compromised campaign. That is what a correct one looks like.

What changes when the advertiser is a clinician

Four things change, and they compound. The first is that the ad platforms treat health as a restricted category, which means some campaign types, some targeting options and some claim language are either unavailable, gated behind certification or verification, or subject to review that ordinary advertisers never encounter. The second is that the data a normal advertiser handles casually becomes sensitive the moment it can be tied to a person and a condition, which changes what may be collected, stored, passed to third parties and used for targeting.

The third is professional accountability. An exaggerated claim on a plumber's ad is a marketing problem. The same exaggeration from a regulated clinician is a matter for the body that issues the license, and the complaint does not need to come from a patient — a competitor screenshotting your ad is enough. The fourth is the one people underestimate: the audience is often anxious, frequently in discomfort, and rarely comparison-shopping in the way a buyer of any other service does. Copy that reads as pushy in a normal vertical reads as predatory here, and it converts worse as well as being wrong.

The practical consequence is that the usual optimization instincts have to be re-ordered. Broadening keywords, expanding radius, adding audience layers and sharpening claims are the four moves a performance marketer reaches for when volume is low. In healthcare, three of the four are either restricted or actively harmful, and the fourth — radius — is the one that most often should be tightened rather than widened.

Query intent in healthcare — the section that decides the account

This is where clinic accounts are won or lost, and it is the part most agencies get wrong because the keyword tools reward getting it wrong. Healthcare search splits into four distinct intents, and they sit at completely different distances from a booked appointment. Treating them as one keyword list is the root cause of almost every clinic account that reports rising enquiries and flat revenue.

Symptom searches

Someone types what they are experiencing. This is by far the largest bucket by volume, it is usually the cheapest per click, and it almost never books. The reason is structural rather than a matter of poor targeting: a person searching a symptom is at the beginning of trying to understand what is happening to them. They do not yet know whether they need a clinician at all, let alone which kind, let alone which one. Many of them are searching on behalf of somebody else. Many are nowhere near you. A large share are not looking to spend money on anything.

Symptom volume is seductive precisely because it is cheap. A clinic account that bids on symptom terms will show impressive impression and click numbers, a low cost per click, and a cost per enquiry that looks respectable if the enquiry definition is loose enough. It will also produce a schedule that does not fill. Symptom queries are content territory — worth ranking for, worth writing about carefully and accurately, worth being the clinic that explains things honestly — and they are almost never worth paid budget.

Condition-research searches

One step along: the person now has a name for the thing, or thinks they do, and is researching it. What it is, what causes it, what the options are, what happens if it is left alone. Intent is higher than a raw symptom search but still overwhelmingly informational, and the same problem applies — you are paying for an education that may or may not end with a booking, may end with a booking somewhere else, and may end with a booking eighteen months from now.

There is one exception worth naming. Where a specific treatment is genuinely elective, high-consideration and the research phase is short, condition research can sit close enough to booking to be worth a small, tightly-matched, separately-budgeted test. That is a test with a defined stopping point, judged on attended appointments, not a reason to open the category.

Treatment searches

Now the person is searching for the thing that would be done, not the thing they have. This is a materially different intent: naming a procedure or a treatment implies they have already accepted that a clinician is involved. On its own it is still often research, but combined with a location it becomes one of the two most valuable query shapes a clinic can buy.

Provider searches

The person is looking for someone to see. A specialty plus a place. A clinic type plus a neighborhood. A named practice. This is the closest thing to booking intent that exists in healthcare search, and it is where the budget belongs. It is more competitive and more expensive per click than everything above it, and it is still the better buy, because the arithmetic that matters is cost per attended appointment rather than cost per click.

Provider searches also include the queries on your own practice name and on the names of clinicians who work there. Whether to bid on your own name is a genuine judgment call with an honest answer on both sides — it is defensive spend on traffic you might have had for free, but it is also the cheapest traffic in the account and the one place where you fully control the message. If competitors are appearing above your own name, the question answers itself.

Getting this split right depends almost entirely on negatives, because broad and phrase matching will pull symptom and research traffic into a treatment campaign continuously and without announcement. The mechanics of building and maintaining that list are the same as for any service business — the companion post on negative keyword lists for service businesses covers the structure and the review cadence, and in a clinic account the list needs reviewing more often than in any other vertical because the informational vocabulary around a condition is effectively unbounded.

Local intent and the discipline of a small radius

For most clinics the catchment is small, and geo-targeting discipline matters more than anything else in the account. People travel a certain distance for a general appointment and a meaningfully greater distance for a specialist procedure, and the difference between those two radii should be visible in your campaign settings rather than averaged into one number.

The failure here is mundane and extremely common: the account targets a whole city or a whole region because that is the default, and a large share of the budget is spent on people who will never travel to you. That spend produces enquiries — people do enquire from far away — and those enquiries produce a cost-per-lead figure that looks fine and an attendance rate that quietly does not.

• Set the radius from where patients actually come from, not from where you would like them to come from. Your practice management system already knows the answer; nobody has looked.
• Split radii by service where travel behavior genuinely differs. A routine appointment and a specialist procedure are different campaigns with different geography and different acceptable costs.
• Check the location-targeting setting itself, not just the radius. The distinction between people in a location and people showing interest in a location is the single most common cause of a clinic paying for traffic on another continent.
• Exclude locations you cannot serve, including areas where a different clinic of your own group is closer to the patient. Bidding against yourself is a real and avoidable expense in multi-site practices.
• Look at the location report by attended appointments rather than by enquiries. Distance suppresses attendance long before it suppresses enquiry volume.

Two adjacent assets do more for local clinic visibility than any bid adjustment: a complete and accurate business profile on the map surface, and consistent practice details everywhere they appear. Neither is glamorous and both outrank most of what an ad account can do for a nearby searcher.

Health information, remarketing, and the line you do not cross

This section is the one to read twice, because it is where a clinic account can cause real harm rather than merely waste money.

The mechanism to understand is that an audience list is itself information. If you build a remarketing audience from everyone who visited the page about a particular diagnosis, procedure or treatment, you have created a list whose membership reveals something sensitive about identifiable people. That is true regardless of whether any name is attached, because the list is used to reach those specific devices and those specific people, and it can be exported, joined, mishandled or subpoenaed like any other data.

The consequences run on two separate tracks and both apply. Ad platforms restrict health-related audience targeting and interest categories, which means the practice can get an account restricted or suspended. Independently of any platform policy, patient-privacy obligations apply to information about a person's health, and a website visit to a condition-specific page can be exactly that. Satisfying the platform does not satisfy the obligation, and the obligation is the more serious of the two.

The knock-on effects are worth stating clearly rather than discovering later. Tracking on condition-specific pages needs deliberate design, not the default install — what fires, what is in the page path, what ends up in a URL parameter, and what a third-party tag can read are all decisions that need making rather than inheriting. Page URLs and page titles are the most commonly overlooked leak, because a descriptive URL naming a condition is transmitted to every tag on the page by default. Form field names travel too. So does anything a chat widget captures.

None of this means a clinic cannot measure its advertising. It means measurement has to be designed around a boundary rather than assuming there is not one. The implementation mechanics — the data layer, what each tag is allowed to see, and how to keep identifiers out of places they should not be — are covered in the conversion tracking implementation guide, and in a clinic account the section on deciding the architecture before touching a tag is not optional preparation. It is the part that keeps you out of trouble.

One more restraint that is easy to skip. Consent handling in healthcare should be more conservative than whatever your local minimum is, because the cost of being wrong is not a fine on a marketing budget — it is a patient discovering that their visit to a page about their condition was shared with an advertising platform. Design for the version of this you would be comfortable explaining to that patient.

What you may and may not say

Ad copy is where clinics get into trouble fastest, and the trouble arrives from two directions at once. The platform disapproves the ad, which is a nuisance. The professional body receives a complaint, which is not.

The categories below are the reliable triggers. They are not an exhaustive list and they are not a substitute for your own regulator's rules, which you must read.

• Outcome guarantees. Any construction promising a result — guaranteed, permanent, risk-free, painless — is both a policy problem and a clinical misrepresentation, because outcomes vary between patients and everyone in healthcare knows it.
• Before-and-after imagery and results claims. Commonly restricted outright, and where permitted usually subject to conditions about representativeness and disclosure that ad formats are badly suited to carrying.
• Superlatives. Best, leading, top-rated, number one, safest. Unverifiable comparative claims about clinical quality are the classic complaint trigger, and they add nothing a patient believes anyway.
• Cure and treatment-efficacy language. Implying a condition will be resolved is a claim about a health outcome and should not appear in an advertisement.
• Urgency and pressure. Limited slots, act now, offer ends. Manufactured scarcity applied to someone's health reads exactly as badly as it should.
• Testimonials, depending on your jurisdiction and profession. Patient testimonials in advertising are restricted for some clinicians and prohibited for others. Check before using any, including ones a patient volunteered.

What is left is plainer and works better than most clinicians expect. Who you are, what you treat, where you are, your qualifications stated accurately, how soon someone can be seen, and what it will cost. That is a complete and genuinely persuasive ad in this category, because it answers the questions a person actually has when they are choosing a clinician while feeling unwell.

The measurement problem

Here is the structural issue that makes healthcare harder to measure than almost any other local service. The conversion that matters is a booked and attended appointment. That event happens in a practice management system, days or weeks after the click, and in the overwhelming majority of clinics that system has never been connected to the ad account in any form. So the account optimizes toward the only things it can see — a form submission, a click on a phone number, a view of a booking page — all of which are several steps upstream of anything that pays for the clinic.

The gap between those proxies and reality is not small and it is not random. Enquiries from far away book less. Enquiries about the wrong service book less. Enquiries generated by vague ad copy book less. Every one of those is a systematic bias, which means an account optimized on enquiry volume drifts steadily toward the cheapest and least qualified traffic while every number on the dashboard improves.

Offline conversion import is the right answer

The correct fix is to send the outcome back. When an enquiry is created, capture the click identifier that came with the visit and store it alongside the enquiry record. Later, when the appointment is booked and then attended, send that status back into the ad account against the stored identifier. The ad platform now learns which clicks produced real appointments and can optimize toward those instead of toward form fills.

This has to be handled carefully in healthcare, and the boundary needs to be explicit rather than assumed, because the data on the other side of it is health information.

That constraint is less limiting than it sounds. The ad platform does not need to know what the appointment was for; it needs to know that the click led to one. If you genuinely need service-level breakdowns — and most clinics do, because acceptable cost differs enormously between a routine appointment and a specialist procedure — get that separation from campaign structure rather than from imported data. One campaign per service line means the service is implied by the campaign, and nothing about any individual patient has to travel anywhere.

Before any of that is built, the basics have to be right, and they usually are not. If the existing conversion actions double-count, fire on page load, or count a booking-page view as a booking, importing appointments on top will produce a beautifully engineered wrong answer. The four-step tracking audit is the right thing to run first, and in a clinic account it is worth running before the first ad goes live rather than after the first disappointing month.

Call handling is the largest single source of waste

Most clinic enquiries are phone calls. People who are unwell, anxious or in pain call; they do not fill in a form and wait. Which means the most consequential part of a clinic's advertising performance happens at a desk, and an unanswered phone is the single largest source of wasted clinic ad spend in this category. It beats bad keywords, bad copy and bad landing pages, and it beats them by a distance, because a missed call is a fully-paid-for enquiry that generates no record of its own existence.

Two things follow. The first is scheduling: run ads when someone answers. Not when the practice is open — when someone actually picks up. Lunch cover, the appointment-heavy hours when reception is with patients, and the period after the last staff member leaves are all windows where the phone effectively does not work, and paying for clicks into them is paying for nothing. If a clinic cannot answer during a window and will not resource it, the correct move is to stop advertising in that window and say so plainly to the practice owner.

The second is measurement, and here the answer depends on where you are. Google's own call reporting and forwarding numbers are only available in a published list of countries, and Nepal is not on it. A call asset — the phone number on the ad — can still be added, but forwarding numbers, call reporting and call-duration conversions are unavailable, so calls generated by Google Ads in Nepal cannot be measured natively. Nepali clinics therefore need either a third-party call-tracking provider or a disciplined manual process, and the manual version is more workable than it sounds: reception asks every caller how they found the practice and records the answer in the same field every time, without exception.

Manual attribution is imperfect. It is also enormously better than the alternative, which is judging a clinic account on the minority of enquiries that arrived by form. If calls are the majority of your enquiries and you measure none of them, you are not measuring your advertising.

• Log every enquiry in one place regardless of channel, with a source field that is filled in every time rather than when someone remembers.
• Measure the answer rate, not just the call count. The number that matters is the share of ad-driven calls a human actually picked up.
• Have a defined path for missed calls — a same-day callback from a list that someone owns — and check that the list is worked rather than assuming it is.
• Give reception a short, written way to answer the questions callers actually ask: earliest availability, cost, location and parking, and what to bring. Uncertainty at this point loses bookings that the advertising already paid for.
• Treat after-hours volume as a decision, not a discovery. Either resource it, route it somewhere, or do not buy it.

Enquiries that do arrive by form need an owner and a response time, and in a multi-clinician practice the routing question is real. The companion post on building a lead routing system that does not drop leads covers the fallback logic, and in a clinic the stakes are higher than in most businesses because the person waiting for a reply is waiting to be seen about their health.

What a clinic landing page actually needs to answer

A patient arriving from an ad is answering five questions, in roughly this order, and the page either answers them in the first screen or loses them. This is a narrower and more human list than the standard landing-page checklist, and it is worth designing directly against.

1. Is this the right kind of clinician for my problem? Say what you treat in the words a patient would use, not in the words on the referral form.
2. Are you near me? Location, visible without scrolling, with the practical detail — parking, the nearest landmark, which floor, whether the entrance is accessible.
3. Do you actually treat my problem? A specific service page beats a general practice page every time, because the person came in on a specific query and is scanning for their own words.
4. How do I book, right now? One obvious path, with the phone number as a live link on mobile and an online booking option if you genuinely have one. Two competing calls to action halve the clarity.
5. What will it cost, and how soon can I be seen? These are the two questions clinics most often refuse to answer on the page, and they are the two that determine whether the person calls you or the next result.

On price: a range with the conditions attached is far better than silence. Silence does not remove the question, it just moves the answer to a phone call the patient may not make. The clinic that publishes an honest range gets fewer enquiries and more bookings, which is the trade every clinic should want and almost none volunteer for.

On availability: earliest availability is the strongest single line of copy available to most clinics, because a person deciding between two comparable practices while feeling unwell will choose the one that can see them sooner nearly every time. It is also honest, verifiable and unavailable to competitors who are busier than you.

The rest of the page mechanics are the same as anywhere else, and worth reading separately — the landing page guide covers message match, form design and mobile in detail. The clinic-specific overlay is restraint: fewer claims, no urgency devices, and a form that asks for the minimum needed to make contact rather than a clinical intake questionnaire that collects sensitive information before the patient has even chosen you.

Reputation is the real conversion lever

In almost every clinic account, the largest available improvement is not in the account. A patient choosing a clinician checks reviews, and they check them at the exact moment between clicking the ad and picking up the phone. Two practices with identical ads, identical pages and identical budgets will convert very differently based on what a search for their name returns, and no amount of bid management closes that gap.

Which makes review generation a paid-media activity in effect if not in name. The process is unglamorous and works: ask every patient, at the same point in their visit, in the same way, with a link that takes two taps. Consistency matters more than cleverness, because a steady trickle of recent reviews reads as a working practice while a cluster of ten from one week reads as a campaign.

Responding to negative reviews needs its own caution in healthcare that does not apply elsewhere. A restaurant can rebut a bad review with the facts. A clinician cannot, because the facts are confidential and confirming that someone was a patient may itself be a disclosure. The safe response is brief, non-specific, does not confirm or deny that the person attended, and moves the conversation to a private channel. Write that response once, agree it, and keep it — the version composed in irritation on the day is the one that causes the second problem.

The metric that lies

Enquiry volume. It is the number on every clinic report, it is the number the account optimizes toward by default, and it can rise steadily for months while the schedule stays exactly as empty as it was.

The mechanism is simple and it is not a measurement artifact. Broaden the keywords toward symptom and research terms, widen the radius, soften the copy so it invites contact from anyone at all, and enquiries go up while cost per enquiry goes down. Every one of those changes also lowers the share of enquiries that become an attended appointment. The two effects are the same effect, and only one of them is on the report.

The second number that catches it is attended appointments, and the ratio between the two is the account's real health. Watch the ratio, not the top line. If enquiries rise and the ratio falls, you have not improved anything — you have bought worse traffic and hidden the fact behind a bigger number. The same discipline applies to how bid changes get made in the first place; the bid governance framework covers making changes against a decided metric rather than against whatever moved this week, and in a clinic account the decided metric should almost always be cost per attended appointment.

Where clinic accounts break

• Symptom keywords in a booking campaign. Cheap traffic, high volume, near-zero attendance. This is the default failure and it is usually visible within a week of looking at the search terms report.
• A radius set by ambition rather than by where patients come from. Produces enquiries from people who were never going to travel, and a cost-per-enquiry figure that conceals it.
• Condition-specific remarketing audiences, built without anyone stopping to ask what the list actually is. Serious, and the kind of problem that does not stay a marketing problem.
• Claims copied from a competitor's ad on the assumption that if they are running it, it must be allowed. It may not be allowed for them either — it may simply not have been reported yet.
• One landing page for a practice with several service lines. The person searched for one specific thing and arrived at a page about everything.
• Calls unmeasured, so the account is judged on form fills that represent the minority of enquiries and the least urgent patients.
• Ads running in hours when nobody answers the phone, most often lunch and the hour after closing.
• Optimizing toward an enquiry conversion action while nobody in the practice can say how many of those enquiries were ever seen.
• A sensitive intake form on the landing page, collecting information from prospects who have not become patients, because someone reused the in-practice form.

When a clinic should not advertise

The honest scope. There are four situations where the right advice is to spend nothing, and a consultant who cannot say so is not worth hiring.

The first is when the schedule is already full. If the constraint is clinical capacity rather than demand, advertising converts a full schedule into a full schedule plus a queue of frustrated people who could not get in, and it damages the reputation that was filling the schedule in the first place. Add capacity, raise prices, or change the mix — do not buy demand you cannot serve.

The second is when the phone is not reliably answered. Everything above compounds here: paid traffic into an unanswered phone is the purest form of waste available in this category, and it cannot be fixed inside the ad account. Fix reception first.

The third is when the practice has no presence on the local map surface and no reviews. Ads will bring people who then search your name and find nothing, and the clinic pays for a click that ends in a comparison it loses. The map listing and the first genuine reviews come first; they are cheaper and they work harder.

The fourth is when nobody can answer whether an enquiry became an attended appointment. Without that loop you cannot distinguish a good month from a lucky one, and you will optimize confidently in the wrong direction for as long as the budget lasts. Build the loop, then buy the traffic.

A starting sequence

If you are setting up a clinic account from scratch, or rebuilding one that is not working, this is the order. It deliberately front-loads the parts that are not advertising, because those are what determine whether the advertising can work at all.

1. Confirm what you are permitted to claim. Read your regulator's and professional body's advertising rules, and confirm your position with a lawyer if any of it is ambiguous. Everything downstream depends on this and nothing else in the account can compensate for getting it wrong.
2. Confirm your privacy position on tracking. Decide what may be collected on condition-specific pages, what tags are allowed to read, and what may cross into any ad platform. Write it down. This is the decision that keeps the account out of serious trouble.
3. Fix the phone. Establish who answers, during which hours, what happens to missed calls, and who owns the callback list. Do this before the first click is bought.
4. Complete the local business profile and start asking every patient for a review, the same way, every time. No incentives, ever.
5. Define the conversion you actually care about — a booked and attended appointment — and find out where that status lives in the practice management system and who can export it.
6. Build enquiry tracking properly: one enquiry record per contact regardless of channel, with the click identifier stored against it and a source field that is always filled.
7. Start with provider searches only. Your specialty plus your area, your clinic type plus your neighborhood, your practice name if competitors are showing on it. Nothing else in campaign one.
8. Set the radius from where your existing patients actually live, split by service line if travel behavior differs. Verify the location-targeting setting itself, not just the radius.
9. Build one landing page per service line that answers the five patient questions above, including a price range and earliest availability.
10. Write plain ad copy. Who you are, what you treat, where, how soon, what it costs. No superlatives, no guarantees, no urgency devices.
11. Run it long enough to accumulate real appointment outcomes, then import those outcomes back against the stored click identifiers — status only, no clinical detail.
12. Only now expand: treatment-plus-location terms first, judged on cost per attended appointment. Add nothing that cannot be judged that way.

Where to start

If you run a clinic and something in this post applies, start with the phone and the reviews rather than the ad account. Both are cheaper, both work whether or not you ever advertise, and both determine the ceiling on everything the account could achieve. Then build the appointment loop, so that when you do start spending you are measuring the thing that pays for the practice rather than the thing that is easy to count.

And keep the order of authority straight. Your regulator and your professional body set the boundary; the ad platform sets a second, different boundary; and a marketer works inside whichever is tighter. Anyone who tells you a claim is fine because the ad was approved has misunderstood which of those documents matters more. I write about performance marketing systems here and build them through Arcetis, the growth systems practice I run, and healthcare is the clearest example of a principle that holds everywhere: the constraint you cannot argue with is usually the thing that tells you what the campaign should have been.

Related reading