Case study
One country, many local markets, one campaign programme
PhysioClick sells a local service nationwide. Physiotherapy is chosen by proximity, so a national campaign is really dozens of local markets wearing a trench coat. This is how we built it city by city, and the measurement problem we had to solve before any of it counted.
What the account actually looked like
- 12
- overlapping conversion tags all active in one Google Ads account
- 24
- tags still firing into Universal Analytics a product Google had already retired
- Dozens
- local markets in one country physiotherapy is chosen by proximity
The problem with "national"
Someone in Haifa searching for physiotherapy is not a prospect for a clinic in Beer Sheva. Yet the default way to run this is one national campaign, one budget, one bid, and a hope that the platform sorts it out. It does not. Cities differ in competition, click prices, clinic capacity, and how often a search turns into a booked appointment, and a single campaign averages all of that into one number that describes nowhere in particular.
Built city by city
We built the account around location, with city-level targeting and copy that names the place a person is actually searching for, then managed it as a single programme rather than dozens of orphan campaigns. That split makes the differences visible: which cities convert, which are expensive for the wrong reasons, and where an extra shekel of budget produces an extra appointment instead of an extra click. Search terms are reviewed continuously, because local intent is where irrelevant queries hide most easily.
Twelve conversion tags counting the same thing
When we audited the measurement setup we found more than twelve active, overlapping Google Ads conversion tags, alongside twenty-four tags still firing for Universal Analytics, a product Google had already retired. The consequence is not cosmetic. A single booking could be counted several times over, so the reported numbers were inflated, inconsistent between sources, and, worst of all, were the numbers the bidding algorithm was being trained on.
Why that had to be fixed first
Smart Bidding is obedient, not wise. It optimises toward whatever you have defined as a conversion, so if the definition is duplicated it learns, diligently, to buy more of the traffic that produces duplicates. Every subsequent decision, which campaign to fund, which city is working, which keyword to cut, inherits that distortion.
None of this was anyone's fault. It is what happens to a website that has been alive for years: each agency, developer, and plugin added a tag, and nobody removed the last one because nobody was quite sure what depended on it. We see it in most accounts we inherit. The work is unglamorous, it produces no immediate chart to celebrate, and it is the single highest-leverage thing you can do to an underperforming account.
How we judge it
By booked appointments in cities where the clinics can actually serve them, not by clicks or by a conversion count we no longer trust. As with every client, we do not publish PhysioClick's commercial performance data. What is fair to say is that this is an ongoing, weekly-managed engagement covering paid search and organic content, and that the measurement now reflects reality.
Frequently asked questions
What does a nationwide local campaign actually mean?
A clinic network sells a local service in many places at once. Someone looking for physiotherapy in Haifa is not served by an ad about a clinic in Beer Sheva, and a single national campaign averages those two very different searches into one bid. So the account is built city by city, with location-specific targeting and copy, while still being managed as one programme with a shared budget and shared learning.
Why not just run one national campaign and let Google optimise?
Because averaging hides the truth. Cities differ in competition, cost per click, clinic capacity, and conversion rate. Merged into one campaign, the strong cities quietly fund the weak ones and you cannot see it happening. Split out, you can move budget toward the places that are actually converting and pause the ones that are not.
You found a tracking problem. What was it?
The account had accumulated more than twelve overlapping Google Ads conversion tags, plus twenty-four tags still firing for Universal Analytics, a product Google retired. Practically, that means the same action could be counted several times, the bidding algorithm was being trained on inflated and inconsistent data, and every optimisation decision made on those numbers was built on sand.
How does that happen?
Gradually, and it is extremely common. Every agency, developer, and plugin that ever touched the site added a tag, and nobody removed the previous one because nobody was certain what would break. It is rarely negligence; it is the natural entropy of a site that has been alive for years.
Do you fix tracking before optimising campaigns?
Always, and it is not a preference. Smart Bidding optimises toward whatever you tell it a conversion is. Feed it duplicates and it learns to buy the traffic that produces duplicates. Fixing measurement first is not housekeeping before the real work; it is the precondition for the real work meaning anything.
Not sure your conversions are real?
If your reported conversions look suspiciously healthy while the phone is quiet, you probably have a counting problem rather than a demand problem. We will audit it and tell you what is actually being measured.