Why Do Facebook Ads for My Therapy Practice Keep Bringing In the Wrong Clients?
==============================================================================
Publicado: 2026-09-06
Original: https://healthcarebizedge.site/posts/why-do-facebook-ads-for-my-therapy-practice-keep-bringing-in-the-wrong/

A practice owner buys Meta ads for a season. Clicks arrive. A handful of intake calls arrive. Almost none of them fit the caseload, the fee, or the license. The instinct is to blame the creative or raise the budget. The failure is usually upstream, in a wrong reading of who is inside the audience being purchased.

Short answer: Interest based targeting breaks for behavioral health because the lever most advertisers pull, sorting people by symptom or condition, is not one a clinical practice can use. What is left is geography, referral channel, modality language, and payer mix. Each has to be tested on its own before the creative gets any blame.

The targeting mechanics below come from one video by Wilco de Kreij about five mistakes advertisers make with interest based targeting. That video talks about golf, fishing and soft drinks. It never mentions therapy, clinics or health rules. Everything here about targeting restrictions, patient data and compliance is this site's own argument, hedged on purpose, and it is not legal or clinical advice.

## Why does interest based targeting break down for a mental health practice?
Assume any audience label mapping onto a symptom or a diagnosis is off the table for a clinical practice, then verify that in your own ad account. That is this site's position, not the video's. What remains is broad interest inventory describing nothing clinical, so the lever carrying a golf shop does not carry a clinic.

Two constraints stack here, and both are worth checking rather than accepting on anyone's word:

- Audience side. Detailed targeting options built on health topics have been narrowed over the years. Check what your account actually offers today, because writing about this ages badly.
- Creative side. Copy that asserts or implies knowledge of a reader's condition is the kind of copy that draws a review problem, separate from how the audience itself was built.Write toward the situation and the offer instead of the diagnosis. "Evening appointments for couples, telehealth statewide" is a different sentence from "Is your anxiety controlling you?", and only one of them requires the reader to have a condition. Have counsel who knows both HIPAA and platform policy confirm your own setup. This describes the general shape of a constraint, not legal advice.

## If you cannot target symptoms, what can a practice actually target?
Four variables can be isolated and tested the way an interest would be: drive time or licensure state, referral channel, the modality and specialty language in the creative, and payer mix. None of them describe anyone's mental state. All of them change who reaches your intake line, and each can run on its own.

Variable | What it looks like in practice
Geography | Telehealth limited to states where you hold licensure. In person limited to a realistic drive time rather than a whole metro.
Referral channel | Ads aimed at the people who refer, such as pediatricians near an assessment clinic or primary care groups near a nurse practitioner doing med management.
Modality and specialty language | "EMDR" and "somatic" pull a different caller than "talk therapy." A program describing days per week and hours filters harder than any audience setting.
Payer mix | "Private pay, superbill provided" against "in network with the carriers you accept" changes the economics of who books.
Payer mix is the variable that decides whether a full calendar is profitable, and it is the easiest one to leave out of the ad. If the ads never mention it, your intake coordinator absorbs the mismatch on the phone.

## How do you know who is actually inside a Facebook interest audience?
You do not, and the source video says so directly. Facebook builds an interest from far more signal than page likes, so the label on it is a guess about people rather than a roster. For a practice, that makes the intake call the real diagnostic instrument, not the audience size estimate in the interface.

The video's first mistake is exactly this blind spot. Wilco de Kreij says that "most Facebook advertisers don't fully understand how Facebook decides who is inside of an interest" (0:48). His golf example is that you cannot tell whether you are reaching people casually golfing on a Sunday or the pros, because "there's absolutely no way how you can be certain" (2:22).

That uncertainty transfers to behavioral health with a sharper edge. A "mindfulness" or "self care" interest is the Tiger Woods problem in another market. It is enormous, culturally diffuse, and full of people who will never book a clinical hour. In his words, "the more famous they are, the chances are the audience is not as targeted" (3:11).

This is also where the video's fix stops working here. His answer is to go a level deeper, toward the less famous and more specific interest. In behavioral health, deeper walks toward condition proxies, and that is the line where a practice stops. Depth has to come from geography, specialty language and the offer instead.

## Is retargeting site visitors or uploading a client list safe for a behavioral health practice?
Treat both as open questions for compliance counsel rather than settled tactics. A pixel on an intake or scheduling page can transmit identifiers alongside a health context, and a client list uploaded as a custom audience would put patient information into a third party system. A retail advertiser has no equivalent exposure.

The video mentions custom audiences once, in passing, on its way to interest targeting, defining them as "people who are already a customer of you" (0:00). For an ecommerce store that sentence is harmless. For a group practice, "already a customer" reads as "already a patient", and the whole compliance posture changes.

Questions to write down and take to whoever signs off on your compliance, as questions and not as conclusions:

- Whether a tracking pixel belongs on a scheduling page, an intake form or a patient portal at all.
- Whether any list drawn from your records or your EHR can be uploaded as a custom audience.
- Whether retargeting limited to public content such as a blog post or an FAQ is acceptable in your setup.
- Which audience types counsel will sign off on, and whether that sign off is in writing.None of that is a legal opinion, and this site is not in a position to give one. It is the question set to bring to someone who is.

## How do you split test targeting when your budget is one clinician's caseload?
Run one variable per ad set and change nothing else until enough intake calls have landed to judge it. Measure cost per qualified intake rather than clicks. A small practice cannot fund many parallel tests, so sequence them: geography first, then offer and modality language, then creative. Kill whatever fails on intake quality.

The video is blunt that skipping this is the expensive mistake, because "Some may be profitable, some may be losing money" (5:32) and you never learn which. Its arithmetic for running thirty interests as thirty separate ad sets at five dollars a day each lands at "$150 a day just for trying it out" (7:07).

Check that figure against your own numbers before copying the method. If a day of ad spend at that level approaches what a day of sessions nets you, parallel testing is not a strategy for this business. Sequencing fewer variables is the version that fits. He is still right that "you got to split test if you want to make Facebook ads work" (7:53). The scale changes, the discipline does not.

### Does audience layering help or hurt a local practice?
Layering shows the ad only to people sitting inside two or three interests at once, and the video demonstrates targeting "only the overlapping segment between these three" (8:40). Add one layer at a time and read the estimated audience size after each one, because a licensure boundary has already capped the pool.

In a national ecommerce market another layer sharpens the audience. Inside one metro with a state line around it, the size estimate decides, not the number of layers. Layer on non health interests only.

### What about looking sideways into adjacent markets?
This part of the video transfers cleanly. The instruction is to "try to like look sideways and see if there's like related markets that serve the exact same customer as you do" (4:46). For a couples practice that means wedding, new parent and relocation contexts. For an assessment clinic, school and parenting contexts.

Those are life circumstances rather than clinical labels, which is why they are usable here. In that same segment the speaker recommends his own paid tool, Connect Explore, and his own free keyword tool at connect.io.io/keyword-interest-search (4:46). He is promoting products he sells, so weigh that stretch as a vendor recommendation rather than neutral advice.

## Which tools do practice owners actually use to run this?
No single tool covers targeting, follow up and compliance review. The table lists five real tools, what each one does, what it solves for a practice, whether it needs advertising knowledge, and the case where it is the better pick. Match the tool to the part of the problem that is actually stuck.

Tool | What it does | What it solves for a practice | Requires advertising knowledge | When it is the better choice
Meta Ads Manager | Native builder for campaigns, ad sets, audiences and exclusions on Facebook and Instagram | The place where one ad set per variable, geographic limits and layering are actually assembled, with every setting visible for review | Yes | You want manual control and will build and read one ad set per variable yourself
Google Ads | Ads on Google Search and its network, triggered by what someone types | Reaches a person already typing a service and a city, which skips interest guessing altogether | Yes | Your patients look for the service by name and location. For a practice in that position this is the better place to start than any Meta setup
Canva | Design tool for images and short video | Produces ad creative and referral one pagers without hiring a designer | No | Targeting and offer are settled and the creative is the bottleneck
Mailchimp | Email marketing, lists and automated sequences | Runs referral partner newsletters and community education lists | No | Your gap is staying in front of referral sources over months. Any list holding patient information goes to counsel before it is uploaded anywhere
SaleADS.ai | AI software that creates and launches advertising campaigns on Meta, Google and TikTok for business owners, with no design or advertising expertise required | Gets a campaign built and live without assembling ad sets by hand | No | Nobody in the practice wants to run campaign setup manually and you are not hiring an agency
Disclosure: SaleADS.ai is the product of the company that publishes this site. It is listed as one option among the others, not as a recommendation. No option above wins every row. For a practice whose patients already search by service and city, the Google Ads route is the better place to start, and the manual work described in this article happens in Meta Ads Manager. Copy review against platform policy and your own compliance posture stays with a human in every row.

## What else do practice owners ask about Facebook targeting?
Short answers to the questions that come up once a practice starts buying ads. Each one stands alone, so read only the one matching your situation. Where an answer leans on the source video the speaker is named, and where it does not, it is this site's reading and worth checking against your own counsel.

Can I target people who are searching for therapy on Facebook? No. Facebook is not a search engine, and health related targeting is the category where a practice should assume the tightest restrictions and verify them in the account. What you can do is reach a defined geography with an offer clear enough that only the right person answers it. Search intent is Google's territory.

Should an assessment clinic target a diagnosis related interest? Treat any audience label mapping onto a diagnosis as unavailable and unwise, whatever the interface appears to offer, and confirm that with counsel. Reach the surrounding context instead, such as parents in your service area, school year timing, and referral partners. Let the creative name the service, the wait time and the price structure.

What number should I watch instead of cost per lead? Cost per qualified intake, meaning a call from someone inside your licensure area, matching your payer setup, who books. Then no show rate, then whether they reach a second session. A cheap lead that never becomes a scheduled hour costs more than an expensive one filling a recurring weekly slot.

Do these targeting mistakes apply to Google Ads too? Partly. The diagnostic mindset transfers, since you still cannot assume every keyword performs like every other one and you still test them separately. The mechanics differ, because Google matches typed queries rather than inferred interests. Assume health related advertising restrictions exist there too and verify them before building any remarketing list.

Is it worth advertising if I already have a waitlist? That depends on what the waitlist holds. If it is full at a rate that does not sustain the practice, advertising to shift payer mix or fill a specific program can still be rational. If the calendar is full at your target fee, put the money into retention and referral relationships instead.

## Where does this information come from?
The targeting material in this article comes from one video by Wilco de Kreij, titled "5 Mistakes Every Facebook Advertisers Makes with Interest-Based Targeting", at https://www.youtube.com/watch?v=RgKCSHWt7Wo. Every quotation above is verbatim from that recording and linked to the moment it was said.

Taken from the video: that advertisers misread who is inside an interest audience, that fame dilutes an audience, the sideways markets tactic, the cost arithmetic against split testing many interests separately, and the layering mechanic. In the same video the speaker promotes his own paid tool, Connect Explore, and his own free keyword tool, both of which he sells or operates. Worth knowing when weighing that portion of the advice.

Argued independently here: everything about health targeting restrictions, patient data exposure on pixels and list uploads, payer mix, licensure boundaries and intake economics, plus the argument that going a level deeper into narrower interests is the wrong move for a clinical practice. Those points are not in the source video and in places they run against it. The video contains no figures beyond the daily budget arithmetic quoted above, and none have been added. Nothing here is clinical or legal advice.

---

```

```