Cardio Clinic
A cardiology clinic in Bucharest · engagement from September 2025
„More patients. That is the whole brief.”The client’s objective, as stated at the start
Over ninety percent of new patients came from one channel. The account was well run. That was exactly the problem. This deck walks through what we did, in the order we did it, and what it produced.
Every figure carries its own window.
Nothing is blended.
| Block | Measurement window | Source |
|---|---|---|
| Business results | Jan 1 to Apr 30, 2026 vs 2025 | client CEO dashboard |
| Google Ads, account totals | Jan 1 to Jul 31, 2026 vs 2025 | Google Ads platform |
| Google Ads, search detail | Jan 1 to May 31, 2026 vs 2025 | campaign export, EUR |
| Branded search | May 1 to Jul 31, 2026 vs 2025 | Search Console API |
| Social reach | Jan 1 to Jul 31, 2026 | social account report |
| Video performance | Sep 2025 to Aug 2026 | TikTok and YouTube analytics |
Search Console keeps sixteen months of history, which is why the branded comparison starts in May. Where a number could be presented two ways, this deck uses the smaller one.
The brief was never
„better marketing”.
It was more patients through the door. A clinic does not need impressions, engagement or a content calendar. It needs someone to walk in on Tuesday morning who was not going to.
- The stated goal. A monthly revenue target that works out to roughly two thousand patients a month.
- Where they were. Around three quarters of the way there, with a gap of about five hundred patients a month.
- Three bottlenecks at once. Acquisition, conversion, retention. Naming which one binds first is the job.
Everything in this deck is downstream of that one sentence. WHY first, then WHAT, then HOW. In that order, every time.
That is where they wanted to go.
This is where they got.
Windows and sources are on the previous slide, and every one of these numbers comes back later with its full table. The rest of this deck is the how, in the order it was actually done.
On the market since 2009.
Excellent medicine. One channel.
The medicine
Around thirty three cardiologists. The only private echocardiography laboratory in Romania accredited by the European Association of Cardiovascular Imaging.
The demand
Over ninety percent of new patients arrived through paid search. No brand demand to speak of. No social presence. One switch, and the whole business runs through it.
A clinic can be the best in the country at what it does and still have a marketing problem. Those are two different assets, and only one of them was built.
More money went in.
The same patients came out.
A well run search account eventually stops being limited by the bidding and starts being limited by the number of people already searching. At that point extra budget buys the same demand at a higher price.
- You cannot bid your way out of it. The pool has a size. Paying more per click does not make the pool bigger.
- You cannot optimise your way out of it either. Better ads inside a fixed pool win share from yourself.
- The only exit is new demand. People who are not searching for the service yet, because they do not know their symptom has a name.
This is the moment where an optimisation brief and a strategy brief stop being the same job.
The obvious move,
and the one we took.
Start at the account
Audit the campaigns, restructure, test keywords, raise the budget, report on cost per conversion. Real work. We did all of it. It was step five.
Start at the phone
Read a full month of inbound calls before writing a single ad. The first growth available was never a new patient. It was the ones already calling.
The order is the method. Do the ad work first and you optimise a message you have not earned the right to write.
We define what the market believes about you
before spending a euro on marketing.
UNRIVALS does not optimise ads. It optimises the speed at which a market understands what you are worth. Everything else is a consequence of that.
- Marketing is how you sell. Most of a category competes here.
- Branding is what you sell. Some of a category competes here.
- Meaning is what it comes to represent for the person on the other side. Almost nobody competes here.
Every competitor sold consultations and investigations. Nobody sold the thing a patient actually wants, which is to stop being afraid.
Two questions.
Two different machines.
What does the market believe about you?
Meaning, difference, proof. Golden Circle, cognitive attribute, content cascade.
Meaning + differentiation + communication + proof = preference.
Where and how do you win profitably?
Market structure, offer, channel, pipeline. Blue Ocean, 4P and 7P, budget cascade.
A healthy brand pushes acquisition cost down, conversion up, lifetime value up.
The second engine is the one clients buy. The first one is the reason it works.
Four layers.
The order runs both ways.
| Layer | What lives there | Who sees it |
|---|---|---|
| L1 · Performance marketing | ads, funnel, traffic | everybody |
| L2 · Revenue and go to market | pipeline, sales process, the path from call to appointment | the operator |
| L3 · AI orchestration | the decision layer, across everything else | almost nobody |
| L4 · Positioning and category | what the brand means and where it sits in the mind | nobody, until it works |
You build upward, L1 then L2 then L4, with L3 optimising at every step. You derive the message downward, L4 sets it and L1 spends against it. A market buys L1 because it is countable. L4 is the reason the counting comes out well.
Seven links.
Skip one and the next one guesses.
- Insight. What is actually true about the customer, from their own words, counted.
- Positioning. What we choose to mean, and to whom.
- Message. The sentence that carries it, written for one person.
- Channel. Where that person already is.
- Execution. The ads, the articles, the videos, the scripts.
- Results. Measured against the business number, not the channel number.
- Recalibration. What the results change about link one.
Most agencies enter at link five and reverse engineer the first four from whatever the ads platform reports.
Seven phases over twelve months.
Cardio Clinic sits in the first five.
| Phase | What it produces | At Cardio Clinic |
|---|---|---|
| F1 · Cognitive Diagnosis | the evidence base | done |
| F2 · Positioning and category | what we mean, to whom | done |
| F3 · Offer and brand architecture | packages, identity, voice | done |
| F4 · Content and go to market | the publishing system | done |
| F5 · Commercial activation | ads, scripts, conversion | done |
| F6 · Performance and advantage | ABM, offer ladder, retention | partial |
| F7 · Strategic recalibration | the next twelve months | not started |
Everything you are about to see came out of the first five phases.
Nine rules.
They decide the order of the work.
- Strategy before optimisation. And infrastructure before marketing.
- Cognitive ownership. Own the space in the mind, and the market comes to you when the need appears.
- WHY before WHAT. One central attribute, carried everywhere.
- Reduce the customer’s mental cost. Every decision you make for them is one they do not have to.
- Marketing and sales are one system. And AI processes the methodology rather than generating filler.
And one gate on everything we hand over: if the client cut our access tomorrow, does the deliverable still stand? If not, we built account management, not strategy.
Before writing one ad,
we read five thousand phone calls.
Source: three rounds of full call analysis · September 2025 (1,367 calls), December 2025 (1,430) and February 2026 (2,287). Transcripts plus metadata, processed with Python and natural language processing. Round one ran to fifteen chapters ending in rewritten scripts and an implementation plan; the later rounds scored every agent individually.
We turned what patients say
into something countable.
| What people actually present with | Share of cases |
|---|---|
| Palpitations | 42% |
| Shortness of breath | 29% |
| Fatigue | 26% |
| Chest pain | 19% |
| Hypertension present in recurring cases | 87% |
Underneath every symptom, one sentence repeated in different words: if something serious is happening to me, will it be found in time. And the finding that reframed the offer: a majority of patients leave reassured rather than diagnosed. The product is not an investigation. It is the end of uncertainty, usually the same day.
What they say on the phone
is not what they type into Google.
Already knows the name of the test. Searching „doppler price”, „holter ekg price”. Comparing on cost. This is who the ads were talking to.
Has a symptom and no idea what it means. Larger, cheaper to reach, and arriving with exactly the fear this clinic is best at ending.
Nothing in the account, the site or the content spoke to population two. Every competitor had the same gap.
Search Console, the account’s own search terms report, People Also Ask and keyword tooling. Two different populations, and only one of them was being served.
Four personas for research.
Three avatars for production.
She manages everyone’s health except her own
Forties to fifties. Postpones her own check-ups indefinitely. Her cardiac symptoms are the ones most often dismissed as stress or tiredness.
His children book the appointment, not him
Sixty plus, managing a condition he already has. Comes on a referral, not on an ad. What he is afraid of is losing his independence and becoming a burden.
He built something and has no time to be ill
Under commercial pressure. Chest tightness filed under stress. Will act on a risk to what he has built, not on a health message.
Personas are research output. Avatars are production tools, and each one gets its own creative set rather than a shared campaign. Mihai was built first, around the check-up package. Elena and Radu came in the second wave, built around the fear instead of the offer.
2019 talked about the machine.
2022 talked about the price.
October 2019. A photograph of an ultrasound probe and an invitation to book. The ad describes what the clinic owns.
February 2022. A magnifier over leg veins and a price in a bubble. The ad describes what the service costs.
Both are competent. Neither is addressed to anybody in particular.
Source: the client’s own banner archive, original file dates October 2019 and February 2022. Thirty two files recovered.
Same clinical truth.
One is addressed to nobody.
Left, 2019: a stock body clutching its chest, and a true statement. Right, 2026: a woman at a laptop at night, in a kitchen full of children’s toys. The subheading names fatigue and shortness of breath, two of the four symptoms the call analysis had counted, and adds dizziness from the same research.
The 2026 headline reads: you take care of everyone, but a heart attack does not ask whether you are ready. Fifteen variants, each mapped to a distinct fear from the research.
Content built as a system,
not as a stream.
- Thirty five articles, built hub and spoke around the symptom clusters the research surfaced, each one internally linked into the cluster.
- Verified medical references on every article. In healthcare the clinic’s credibility is the asset being spent.
- A medical review gate before publication. It slows the calendar down and it is not optional. Nothing ships that a cardiologist has not signed.
- Topics chosen from what patients ask, on the phone and in search, rather than from what an agency finds interesting.
A stream of articles buys traffic. A cluster buys a position on a topic, which keeps paying after you stop publishing.
The video formula came out of the research,
not out of the platform.
Three beats, in this order: symptom, then question, then fear. Does your heart race. Does your blood pressure spike when you are upset. Topics taken from what patients ask on the phone and type into search.
Source: social account reporting, February 2026. Five videos above one million views in that month, seven above eight hundred thousand.
The least glamorous work here,
and the highest return.
The call analysis from step one produced rewritten scripts, per agent scoring, and training. It raises conversion on demand that has already been paid for, which means it costs nothing per additional patient.
- Conversion per caller rose 52 percent year on year, to just under seventy percent.
- Not one extra call was bought to produce it. The calls were already arriving.
- This is what „marketing and sales are one system” means in practice.
An agency that only holds the ad account cannot touch this. Which means it cannot fix the most expensive leak in the business.
The clinic has its own brain.
It stays whether we do or not.
- A structured vault of its research, brand rules, personas, scripts and deliverables. One place, versioned.
- An agent that transcribes and analyses sales calls continuously, so step one stops being a one off study and becomes a running instrument.
- An agent that manages review responses on the Google Business Profile.
- Dashboards, and the analytics feed wired into a data warehouse, so the history stops depending on any one platform’s retention window.
This is layer three, and it is what makes the other three compound. It is also the answer to the gate question. If the engagement ended tomorrow, the clinic keeps the brain, the research, the scripts, the article system and the data.
January to April 2026,
against the same months of 2025.
| Metric | Year on year | What it is |
|---|---|---|
| Revenue | +39.4% | the business result |
| Appointments | +57.6% | real demand |
| Calls | +44.6% | top of the funnel |
| Investigations | +86.9% | higher margin revenue |
| Unique patients | +49.8% | a genuinely new patient base |
| Conversion per caller | +52.0% | operational efficiency |
| Returning patients | +39.2% | retention |
Four months, because that is the closed period in the dashboard. The shortest window in this deck, and the one that understates the year.
Spend went up on search.
So did the click rate. Cost per conversion fell.
| Search campaigns only | 2025 | 2026 | Change |
|---|---|---|---|
| Click-through rate | 12.67% | 13.62% | +0.95 pp |
| Cost per conversion | €0.00 | €0.00 | −6.2% |
| Conversions | 0,000 | 0,000 | +37.7% |
| Spend | €00,000 | €00,000 | +29.2% |
| Branded search, cost per conversion | €0.00 | €0.00 | −22.5% |
Two figures we deliberately do not feature: the blended click rate and the blended conversion rate. Both fell, because awareness inventory was scaled more than fivefold. That is a mix effect, and presenting it either way would be misleading.
Three times the impressions.
A third more money.
| Whole account, all campaign types | Change | What moved it |
|---|---|---|
| Impressions | +204.5% | awareness inventory opened up |
| Interaction rate | +70.2% | creative built from the research |
| Average cost per interaction | −74.8% | the mix moved to cheaper attention |
| Spend | +30.7% | budget grew, but nowhere near reach |
Cost per interaction did not fall because the bidding got clever. It fell because we stopped buying only the expensive end of the market. That is the mix shift, and it was the plan, not a side effect.
Seven months.
From no presence to a recognised name.
Sources: social account report for impressions, January 1 to July 31, 2026, paid split TikTok 15.72M, Meta 11.85M, Google 4.5M. Follower counts pulled from the platform API on 12 August 2026: Facebook 17,093 to 22,525, TikTok 8 to 6,820, Instagram 68 to 354, YouTube 20 to 269. The same API returns nothing at all for 2025, because there was no tracked presence to return.
Twenty nine videos
went past a hundred thousand views.
Every topic came off the phone, not off a trend. Check any of it yourself: tiktok.com/@cardioclinic.bucuresti · youtube.com/@CardioClinicBucuresti · facebook.com/CardioClinicBucuresti. Sort the YouTube shorts by most popular and the eight are the first eight.
Source: TikTok and YouTube analytics for the clinic’s own accounts, September 2025 to August 2026. Views count paid distribution as well as organic, the same way every reach figure in this deck does.
More people typing the name.
Not better ranking for it.
The site was already inside the top three for its own name in both periods, so this is not a ranking gain. Demand for the name grew while visibility for it stayed where it was. That is what brand lift means, and it is the one figure here a sceptic cannot explain away with a platform change.
Source: Google Search Console API, property cardioclinic.ro, pulled 12 August 2026. Both spellings of the name included, thirty five queries belonging to other businesses removed. Sixteen month retention is why the comparison starts in May.
How a brand makes
acquisition cheaper.
- Clarity produces memory.
- Memory produces orientation. They know who you are before they need you.
- Orientation lowers the mental work of choosing.
- Less mental work produces trust faster.
- Trust converts. And every link above takes cost out of the step after it.
The observable version here: branded demand rose sharply while the position for the brand stayed flat, and in the same period the cost per conversion on the paid brand term fell by 22.5 percent. Work on channels with no direct attribution made the channel with direct attribution cheaper.
The ratios got worse.
That was the plan.
The mandate was to break a ceiling, not to spend less. Ratios compress when you double the investment, and new patients arriving from awareness start at a lower value than patients who arrived already convinced. Meanwhile, on the core channel, spend rose 29 percent while the click rate rose and cost per conversion fell. That is the part that says the money was spent well.
Four things we could have claimed
and did not.
- No traffic claim, anywhere. Analytics sessions fell 18.6 percent and organic sessions fell 58 percent between January and May 2026, against the same months of 2025. It may be a tracking migration, a consent change, or a real effect. It is under investigation, so we say nothing about traffic.
- No working media efficiency figure. The revenue dashboard and the ads export do not reconcile cleanly across currencies and periods, so the metric was dropped rather than estimated.
- No causation claim on brand lift. The rise correlates with the period of work. Correlation is what the data supports, so correlation is what is claimed.
- No cherry-picked campaign list. Five of eight search campaigns improved their click rate. Three did not. Both numbers are on the earlier slide.
A case study that only contains good news is a case study a serious buyer discounts entirely.
„Was that you,
or was the clinic already growing?”
It was already growing. In 2025 against 2024, before this engagement began in September 2025, the clinic increased total patients by 146 percent and unique patients by 57 percent, on a marketing budget 70 percent larger. Return on ad spend fell 25 percent in that same year.
- So the direction is not the claim. The clinic was already buying growth with budget, and the ratios were already compressing.
- The claim is about the mechanism. Single channel concentration broken. Branded demand created where there was almost none. Conversion on existing calls raised by 52 percent without buying one extra call.
- Growing a business that is already growing is the harder problem, not the easier one. The easy percentage gains were already taken.
Source: the client’s marketing dashboard, full year 2025 compared to full year 2024. The engagement covers four months of that window.
That result came from
under a fifth of the system.
| Still not deployed | Why it matters |
|---|---|
| Category architecture in the full sense | owning the space, not renting attention in it |
| Offer ladder beyond the existing packages | revenue per patient, not just patient count |
| Retention and lifetime value engine | the metric that fell, addressed directly |
| Account based work on referrers | a second demand source that is not advertising |
| Email, built and blocked | twelve pieces written, held behind a consent problem in the existing contact base |
Five of seven phases executed, and the two that compound hardest are the ones still ahead. What you have just seen is the early part of the method, done properly and in order.
The clinic is not the point.
The pattern is.
One channel that works, at a ceiling, where more budget stops buying more customers. That describes most manufacturers in a price war, most B2B service firms competing on rate cards, and most specialist clinics.
- Read what customers actually say before writing anything.
- Fix conversion on demand you have already paid for. It is the cheapest growth in the business.
- Build recognition on channels that do not report attribution, so the channels that do get cheaper.
- Then scale. In that order, not the reverse.
Growth from the system, not from a bigger number in the same box.
The next moves are already
written on the numbers.
- Lifetime value, directly. It fell 30 percent because the patient base got younger in tenure. Retention is now the highest leverage number in the business.
- Unblock email. Twelve pieces are written and sitting behind a consent problem in a base of roughly six thousand contacts. Re-permission first, then the sequence.
- Resolve the analytics discrepancy. Until sessions reconcile, an entire channel cannot be argued about honestly.
- Then category architecture. The clinic owns a real medical distinction that nobody in the market is claiming out loud.
None of these is a bigger advertising budget.
If the shape of that problem
is familiar, the diagnosis is the place to start.
A clinic locked into one channel, at a growth ceiling. We built a multi channel demand system, and in roughly seven months: revenue up 39 percent, appointments up 57 percent, investigations up 86 percent, through brand and efficiency rather than through a bigger budget on the same channel.
unrivals.ro/cardio-clinic
Every figure in this deck traces to a named source and a stated window. Where two readings were possible, the smaller one is on the slide.
