Case study: a Denver private practice entering a new market
A licensed psychotherapy practice with an established patient base in one state opened a second office in Denver. In the new market it had no referral network, no local reputation, and no inbound demand. The goal was straightforward and hard: fill a schedule in a city where nobody knew the practice existed, weighted toward self-pay work rather than insurance billing.
This is the practice's own account of the marketing, anonymized at my client's preference. It contains no patient information of any kind, because it never could.
What was actually wrong
The practice was already spending on Google Ads and had been for weeks, with almost nothing to show for it. The assumption going in was that the budget was too small. It was not. The diagnosis found five separate problems, and the budget was not among them.
- Wrong campaign type. The account was running a Smart Campaign, which forces multi-network delivery and cannot be opted out of the Display Network. Roughly 70% of spend was going to Display placements, not search.
- A geographic radius that excluded the market. Targeting was set to a seven-mile ring around the office, which cut out most of the metro area and all of the military population an hour south.
- Keyword themes working against intent. The themes included terms describing adjacent services the practice does not provide, which attracted clicks that could never convert.
- A business-hours-only schedule, which misses the evenings and weekends when people actually search for a therapist.
- No conversion tracking wired to the ad account at all. The campaign was optimizing toward clicks because clicks were the only thing it could see.
The problem that was not in the ad account
The most expensive problem was not advertising at all. It was discovered a week after launch, when the site showed phone taps but no new patient conversations were happening.
The local phone number on every page and on the Google listing was ringing a line nobody answered for business. Every single ad click that turned into a phone call had been leaking on the floor. A second issue compounded it: the practice's own phone was auto-blocking calls from the new area code as suspected spam, so even corrected calls were being filtered before anyone heard them ring.
Both were fixed the day they were found. No amount of budget, bidding or copywriting would have surfaced either one. They were only found by tracing what happened to a lead after it left the website, which is the part most people skip.
What was rebuilt
- The Smart Campaign was paused and replaced with a Search-only campaign, with Display and Search Partners switched off, so the budget buys search intent and nothing else.
- Geography was re-drawn to cover the actual metro area plus a second radius around a large military installation, on a presence-only basis.
- Keywords were rebuilt around what a person in distress actually types, with the anti-intent terms removed.
- A dedicated landing page was built for the new market, rather than sending paid traffic to a homepage written for the original state.
- The duplicate Google Business Profile listing, which had a misspelling in the business name and was splitting the practice's local ranking signals, was cleaned up.
- Email forwarding for the practice domain, which had been silently dead for weeks, was restored, along with an automated weekly analytics report so the owner can see performance without logging into anything.
Results
Measured from the practice's Google Analytics 4 property, comparing the four weeks before the rebuilt campaign launched against the three weeks after.
| Metric | 4 weeks before | 3 weeks after | Change |
|---|---|---|---|
| Visitors per week | 23 | 86 | Up 275% |
| Leads per week (phone and email clicks) | 1.75 | 5.3 | Up 205% |
| New self-pay patients | 0 from this channel | 3 | New |
The traffic is the right traffic, which matters more than the volume. In the most recent full week measured, paid search was the single largest channel at 46 of 89 sessions. The new market-specific landing page was the most-visited page on the entire site. Users in the target metro made up 35% of all traffic, against 5% from the practice's original home city. Display spend was zero, down from roughly 70% of budget. Average time on site was 2 minutes 36 seconds with a 42% bounce rate, which is engaged reading, not accidental clicks.
What I am not claiming
This section exists because most case studies leave it out.
- I am not claiming a return on ad spend. I have not published a cost per patient or a revenue multiple, because the exact spend figure for the period has not been pulled from the ad account yet. When a case study leads with a ROAS number, ask how it was calculated.
- I am not claiming the ads caused all three patients. Channel attribution at the traffic level is strong and documented above. Patient-level attribution is not: the practice did not ask each new patient how they found it. Earlier in this same engagement, a new patient who looked like an ad result turned out to have arrived through organic search. That is exactly why I will not assert it here.
- Three patients is a small number. It is a real result in a market that produced none, and it is still three.
- Traffic eased over the three weeks, from 92 to 89 to 78 weekly visitors, while leads held. New campaigns often spike and settle. The trend is worth watching rather than hiding.
What this engagement says about the work
The headline number here is not the traffic lift. It is that the highest-value fix was a phone line, found by following a lead past the point where most reporting stops. Ad platforms will happily report a successful campaign while every call it generates goes unanswered.
If your ads are running and the phone is not ringing, the problem may not be the ads.
Published with the client's permission. No patient information was used, and none ever will be. No reviews or testimonials were solicited from patients, consistent with professional ethics rules for licensed psychologists.
Common questions
Over the three weeks after the rebuilt campaign launched, weekly website traffic rose from 23 to 86 visitors, a 275% increase, and weekly leads rose from 1.75 to 5.3, a 205% increase. The practice took on three new self-pay patients in a market that had produced none. Figures come from the practice's own Google Analytics 4 property.
Because the exact spend for the period has not been pulled from the ad account yet, and I will not publish a multiple I cannot show the arithmetic for. A case study that leads with a ROAS number without showing spend, revenue and the attribution method behind it is asking to be taken on faith.
No, and I say so on the page. Channel-level attribution is well documented: paid search was the largest traffic source and the market-specific landing page was the most visited page on the site. Patient-level attribution was never captured, because the practice did not ask each new patient how they found it. Earlier in the same engagement a patient who looked like an ad result turned out to be organic.
Not anything in the ad account. The local phone number published on the website and the Google listing was ringing a line nobody answered for business, so every call generated by the campaign was being lost. A second problem compounded it: the practice's phone was auto-blocking the new area code as spam. Both were found by tracing what happened to a lead after it left the website.
At their preference. Health-care practices often prefer not to publicize that they advertise, and the results read identically without the name. Every figure above comes from their analytics and is published with their permission.
Yes. SteelBase works with local service businesses and private practices generally. The pattern in this engagement, where the leak was operational rather than in the ad account, is common across industries.