Running Google Ads for an aesthetics practice

Why this category is genuinely different

Most paid search advice assumes a full toolkit: retarget the visitors, upload the customer list, build a lookalike from converters. In a sensitive health category none of that is available, and the difference is structural rather than a matter of degree.

Google's personalized advertising policy defines Health as a sensitive interest category, and the definition explicitly includes invasive medical procedures, including cosmetic surgery, surgical procedures, or injections.

That captures the core of the aesthetics business. It does not matter that treatment is elective, cash-pay, or marketed as aesthetic rather than medical. The policy is written around the procedure being advertised, not around how the business describes itself.

The practical consequence is that a great deal of standard agency advice is simply inapplicable here, and an agency that has not run this category will often not know that until a campaign is built.

The targeting you cannot use

Google splits targeting into advertiser-curated audiences and predefined Google audiences, and applies the restriction to the first group. Advertiser-curated audiences cannot be used when promoting in a sensitive interest category.

Unavailable: Customer Match, your data segments, audience expansion, lookalike segments.

Still available: in-market segments, affinity, demographics and detailed demographics with exceptions, life events, location targeting, custom segments.

So the patient-list upload is out, the lookalike built from people who booked is out, and expanding an audience that performed is out. Remarketing to someone who viewed a treatment page is the same restriction in different clothing, because it is your data segment either way.

Of what remains, custom segments do the most work, because they are assembled from search behaviour rather than from a list of identified people. Location targeting is untouched, which matters disproportionately for a practice whose realistic catchment is a thirty minute drive.

It is worth being blunt that none of this replaces a patient list. The account leans harder on query intent and lighter on audience precision than an unrestricted account, and that is a property of the category rather than a deficiency in the setup. Full detail in what Google's health targeting rules forbid.

Two ways volume disappears without an error

Healthcare accounts have a distinctive failure mode: everything is approved, nothing is flagged, and the campaign serves less than its settings imply. There are two separate mechanisms behind it and neither shows up as a disapproval.

Eligible (limited). Google's healthcare and medicines policy states that some content can only be advertised in certain locations by advertisers who have applied and been approved. Where the campaign targets allowed locations and the domain is properly certified, the ad is labelled Eligible (limited) and runs inside that boundary. This is the normal healthy state for a lot of legitimate healthcare advertising, not a warning.

Limited Ad Serving. Separately, Google caps impressions for advertisers it judges unqualified, weighing account maturity, advertiser industry, verification status, policy history and user reports. Google states that individual ads will not be disapproved. From August 2026 this extends to all Google Ads products, completing by 2028.

Both produce the same symptom: keyword volume exists, budget is not exhausted, impression share is low, and there is no auction explanation. In a healthcare account that combination should send you to policy status before it sends you to bids, because raising a bid on a limited ad buys nothing.

What to count, and why it matters more here

Remove the audience layer and four levers remain: the query, the creative, the landing page, and the conversion you report back. In an unrestricted account a good audience partly compensates for a weak conversion definition. Here nothing compensates.

Whatever you count is what the bidding will go and find more of. That is true everywhere. What is different in this category is that there is no audience constraint to catch the error, so optimizing toward a raw form fill trains the system on the wrong population using real money.

The result is familiar to anyone who has run one of these accounts. Conversions look healthy, cost per lead is flat or improving, and the front desk reports that half the enquiries were price-shopping or never going to book.

Count the booked consultation rather than the request for one. Better still, the consultation attended. Each step further down is a stronger signal and a smaller number, and the tension between those two is the real design decision, because an event firing four times a month cannot steer automated bidding.

The workable compromise is to optimize toward the strongest event with enough volume to be learnable and report on the one below it. That is a deliberate choice with a stated reason, which is different from optimizing to a form fill because it was the default. The mechanism is offline conversion import, and the method is the qualified-lead feedback loop.

Where advantage actually comes from

The restrictions apply to every advertiser in the category equally. That is unusual, and it changes where an edge can exist: not in targeting sophistication, which nobody has, but in operational plumbing, which almost nobody does.

A practice that can tell its ad platform within a day or two which enquiries became booked consultations has an advantage its competitors cannot buy with budget or data, because that data is not available to them either.

In most verticals a targeting restriction is a disadvantage. Here it is a levelling, which turns feeding real outcomes back to the platform from a best practice into the main axis of competition.

That is our reading of what follows from the policy rather than something Google states. What is not an inference is the direction: the audience tools are unavailable, the conversion definition is not, and one of those is therefore where the remaining leverage sits.

The order to do this in

Fix measurement before targeting, and check policy state before bids. Both orderings are the reverse of what most accounts get, and both are cheap.

• Audit what is attached. If a Customer Match list, data segment, audience expansion setting or lookalike is attached to anything promoting a covered procedure, that is the finding, and it usually predates whoever runs the account now.

• Check policy state. Are ads Eligible (limited)? Is there a limited-serving notification? Complete advertiser verification, which is the one qualification signal you directly control.

• Fix the conversion. Get to booked or attended consultations before touching bidding strategy. Everything downstream is denominated in this.

• Rebuild targeting on what is permitted. Custom segments around the terms a considering patient actually uses, plus location and in-market. Not as a replacement for the patient list, as a different instrument.

• Then optimise. Bids, budgets and creative, against a signal that now means something.

This page states what the policies say and how we read them. It is not legal advice, and advertising rules for medical practices also come from state boards and other regulators Google does not administer.

Targeting availability in a sensitive health category

Two ways volume disappears without an error