Results
Three healthcare accounts, and what actually changed in each.
No client is named on this page. The figures are cleared for outside use and the names are not, which is a standing rule here rather than something we negotiate account by account.
Every number below comes from a dated internal report, and it is written in the tense that report was written in, rather than as though it were still true this morning. Each study carries the caveat that belongs to it, printed next to the figure it governs. One of the three opens with its caveat, because the account makes no sense without it.
None of this forecasts anything for your practice. Three accounts across three specialties is not a sample. Two are multi-location groups running at a scale most practices are not, and the third buys clicks in Manhattan. What travels between them is the method. The numbers stay where they came from.
Each is told in four parts: what the account looked like, what we changed, what moved, and what we would do next. The last part is here because all three of these accounts are unfinished.
Case study 01 — urgent care
An urgent care group, from invisible to page one, January 2024 to April 2026
Where the account stood in January 2024
A multi-location urgent care group came to us doing 17,351 organic sessions a month. It held 736 keywords in the top three results and had 1,911 buried past position 50. On the paid side, a booked patient appointment was costing $122.36.
None of that is unusual for a group that grew by opening clinics rather than by marketing them. What we saw when we opened the account was one mismatch repeated at every address: the site had been built around the brand, and the patients were searching for a neighborhood.
What we changed, location by location
We treated every clinic as its own contest. A page per site, carrying that site’s hours, its directions, what it handles and what it sends to an emergency room. A Google profile per site, maintained monthly rather than claimed once. Then we rebuilt the location pages that were functioning as addresses, and split the ad account along the same lines so that each clinic bid against the radius it could realistically serve.
The conversion we optimized the paid account toward was the booked appointment, from month one. An account pointed at clicks will buy you more clicks, cheerfully, for as long as you fund it.
Underneath both channels we built a library: 88 pieces over two years, made up of 70 articles, 15 pillar pages and 3 on-site. Roughly three a month, sustained. An editor went through every one of them before it published, and anything clinical went to the client’s clinicians first.
What twenty-six months moved
17,351 → 62,602
- Top-three keyword rankings
- 736 → 4,743
- Cost per booked appointment
- $122.36 → $8.80
- Appointments booked through ads
- 49,320
Organic became the largest patient channel the group had. Over the same stretch we took total ranking keywords from 2,972 to 9,417 and cut the tail buried past position 50 from 1,911 down to 483.
The paid figure is the one people remember. We cut the cost of a booked appointment from $122.36 to $8.80, a fall of 93%, and did it while the number of appointments was going up rather than down. Across the engagement the ads produced 49,320 booked appointments on 3.1 million impressions.
What we would do next, and the stretch we have to account for
The account finished the period limited by budget rather than by performance. That is a rare place for an account to be and it is the first thing we would act on. At $8.80, another $1,000 a month buys roughly 114 more appointments. The ROI calculator runs that same arithmetic against your own cost per appointment, which is worth doing before anyone quotes you a budget.
There was also a softer stretch in 2026 that the April review does not cover. Nobody has re-pulled the numbers since, so this page publishes the two-year arc and quotes no current-month figure. We would rather show you a dated number than a fresh-looking one we have not checked.
Case study 02 — behavioral health
A behavioral health network converting at nearly double the rate, on less than half the traffic
Start with the reset, because every figure below sits on it
A 17-location behavioral health network came out of a site migration with its organic traffic reset. Year over year, clicks were down 57% to around 95,600 and impressions down 31.5% to roughly 10.9 million. Admissions were down 18%. Some of that is the migration itself, and some of it is a branded paid campaign buying terms the network already ranked first for organically.
That is the account. Every win below happened inside it, and a page that quoted the wins without the reset would be describing a different company. What a rebuild does to a site that already ranks has a page of its own, because it is the most common way a practice loses what it spent years earning.
The problem underneath the reset
The traffic and the demand were sitting on different pages. Blog traffic converted at 0.3%. Facility pages converted at 20.8%. The blog was most of the visits and almost none of the inquiries.
Nobody selling content volume goes looking for that finding. We saw it only because the admissions data was open next to the analytics, and most reporting in this category stops several steps upstream of an admission.
What we changed about which pages get built
We stopped adding to the blog. What we built instead were facility and service-area pages, one per location, each written to answer what somebody deciding at eleven at night actually needs to know. Seventeen Google profiles went onto a maintenance cadence. The commercial near-me searches became the target in place of the informational ones.
We also put the admissions numbers into the marketing report, which is what made any of the rest of this arguable.
What changed by May 2026
3.43%
- Facility pages
- 8% of visits, 50% of leads
- Facility pages vs blog
- 20.8% vs 0.3%
- Profile views, 17 listings
- 48,557
The commercial near-me terms climbed in the same window. Between April and May, “alcohol rehab near me” went from 18 to 5. “Mental health treatment near me” went from 13 to 2. “Inpatient mental health facilities near me” came up to 3 from 27. The network picked up 39 new top-three rankings that month and its Domain Rating reached 65. Across the seventeen listings we counted 48,557 profile views, with direction requests up 10.4% on the month before.
Then admissions turned. January through May ran 261, 209, 206, 223, and 227. That last figure is still a dip from January and it is the second straight month of recovery out of the low. May also came at the year’s best lead-to-admit rate, 6%. Fewer inquiries, better ones.
What we would do next
The volume has a recovery to finish, and that is the open item. Conversion cannot keep doubling. At some point the traffic has to come back, and getting it back means finishing the technical work the migration interrupted rather than publishing more.
The other thing we would look at is the branded paid campaign. If it is buying clicks on terms the network already owns organically, some of that budget is being spent to move a patient from one of its own results to another. Proving it either way needs a holdout test, not a meeting.
These are one month’s figures from a May 2026 report, and we have not re-pulled them since.
Case study 03 — dermatology
A Manhattan dermatology practice that cut its cost per lead 45% in three months
Where the account stood in February 2026
A Manhattan dermatology practice was paying $19.80 for a lead in February 2026. Medical and cosmetic dermatology were running through one account, which averages two different buyers into one price and hides which of them is the expensive one.
Before any of the numbers: these are leads, not booked patients. The practice has no lead-to-client tracking connected, so every figure in this study measures media efficiency and not what a patient costs to acquire. The report we wrote says so in its own appendix. It is also the reason this study ends where it does.
What we changed inside the ad account
We split the account along the lines that move the price rather than the lines that describe the services. Geography was one of them, and language turned out to be the other, which is not the split anybody involved expected to matter most.
Then we spent three months moving budget toward whatever was returning a lead cheaply and away from whatever was not. That is most of what paid search actually is: the thinking sits in the structure of the account, and the work sits in the reallocation.
What three months moved, and what the number does not say
$19.80 → $10.86
- Leads on $6,495 of spend
- 628
- Blended cost per lead
- $10.34
- Spanish ad group click-through
- 9.2% vs 4.3%
Between February 15 and May 20, 2026 the account produced 628 leads on $6,495 of spend, a blended $10.34 each. March was the strongest month: 248 leads at $7.85. The period converted at 19.8% on 3,175 clicks, with a 4.7% click-through rate and an average cost per click of $2.05.
The Spanish-language ad group was the best thing in the account. It returned 124 leads at $9.68 apiece on a 9.2% click-through rate, against 4.3% for the equivalent English ads, while carrying the second-largest spend of any group in the account. The bilingual audience was the underserved one, and nobody was bidding hard against us for it.
Geography concentrated harder than anyone had assumed. When we measured the leads by postal code, a single ZIP had produced 499 of the 628, which is nearly four in five.
Two things those numbers do not say. April’s cost per lead rose 27% over March on flat spend before partly recovering in May, which is why this study quotes a period and not a month. And $10.34 is not a cost per patient. Nobody working on that account can tell you what a patient cost, ourselves included.
What we would do next
Connect the down-funnel tracking. That has been the defined next step since the report went out. Until a form submission can be traced to a booked appointment, the account is being optimized toward the cheapest lead rather than the most valuable one, and those are not always the same campaign. What that work involves is set out on its own page, along with what it still cannot reach.
The second thing is that ZIP code. Seventy-nine percent of leads out of one area is either a strength to press or a ceiling to break, and finding out which one it is takes a test rather than an opinion.
What the three accounts have in common, and what they don’t
Three habits ran through all three accounts, and between them they are worth more than any single figure above.
We connected the measurement before we spent anything. In the urgent care group the booked appointment was the conversion from month one. In the behavioral health network the admissions numbers went into the marketing report. In the dermatology account that step never happened, and the study above is what that costs you.
We built the page types that convert instead of the ones that are easy to produce. Facility pages at 20.8% against a blog at 0.3% is the cleanest version of that argument we have, and it is why the content plan on every account starts with an argument about page types.
And we measured media against the thing the practice actually counts rather than against the thing the platform reports. An appointment for the urgent care group. An admission for the network. For the dermatology practice a lead, because that is as far as the tracking reaches today.
That is not a system and it does not have a name. It is three habits, applied by people with the data open in front of them, and in the accounts we run they are most of the difference between a report that explains something and a report that only reassures you. What did not repeat matters too. The three accounts are different sizes in different specialties, two of them had a problem the third did not have at all, and the urgent care arc took twenty-six months while the dermatology arc took three. Each piece of this has a page saying what it involves and what it does not do.
One note on what is missing here. Michael ran the marketing for a national wound care practice and for a medical-ID nonprofit before founding Dweeso, and both went well, but neither was a Dweeso engagement, so they sit on the about page as his own track record rather than here.
Questions
Questions buyers ask about these numbers
- Why isn't a single client named on this page?
- The figures are cleared for outside use. The names are not. That is a standing rule here, and it costs us the logo grid every competing page has. It does not cost you anything real: the numbers below are exact, dated, and taken from the reports we sent the clients. An exact figure under a generic description tells you more than a round one under a famous name.
- Are these results typical?
- No, and we would not know. Three accounts is not a sample. Two of them are multi-location groups with budgets most practices do not have, and the third buys clicks in Manhattan, where a click costs nothing like it costs in Ohio. What transfers between them is the method. The numbers stay with the accounts they came from.
- Why does every figure here carry a caveat?
- Because the reports do. The behavioral health engagement is a conversion turnaround inside a traffic reset, and quoting the conversion rate without the reset would describe a different company. The dermatology figures are leads rather than patients, because nothing connects the form to the chair yet. Take either caveat away and the number stops being true.
- How long did each of these take?
- They are three different durations, which is most of the answer. The urgent care engagement began in January 2024 and the figures here close in April 2026, so twenty-six months. The dermatology account is three months, February 15 to May 20, 2026, which is how fast a paid account can move. The behavioral health figures come from a May 2026 report, with the admissions arc traced back to January.
See what your own numbers look like.
Tell us the practice and where it operates. We will send back what it looks like in search today, who is holding the results it should hold, and what we think is in the way. It comes in writing, and nobody has to sign anything to get it.