A few weeks ago I was on an audit call with a client's marketing team. I worked through their Google Ads account line by line, and when we got to the cost-per-conversion on their biggest campaign, the call went quiet until one word was uttered: "Insane."
They weren’t wrong. That campaign cost just over $9,400 over the course of thirty days, and resulted in less than six conversions. When we looked closer at them, only three were genuine new-patient enquiries. The rest were people clicking "get directions," and one person who downloaded a price guide.
That means each real enquiry cost $3,100.
This didn’t happen because of weak ad copy, a bad offer or just poor luck that month, it was because there were two campaigns from one account fighting over the same vocabulary, with nothing in place to tell them apart.
The Setup: Two Clinics, One Shared Vocabulary
This client is a dental group. Two sites with the same owner in the same city providing two genuinely different services.
The main clinic does general dentistry like check-ups, hygiene, fillings and crowns. A few suburbs away they have a separate specialist orthodontic clinic doing braces, clear aligners and complex bite work.
Different practitioners in different places, with different websites and separate books.
There were two ad campaigns between them: a larger general dentistry campaign running around $9,400 a month, and a smaller orthodontics campaign running closer to $1,200.
On paper that is a tidy structure. In action it was anything but.
The problem is that a prospective client doesn’t know or care about the org chart. Someone searching for "orthodontist near me" and someone searching for "dentist near me" are, to Google, asking more or less the same thing, as both queries contain the shape of "a person who fixes teeth, close to me, taking new patients."
Problem One: Match Types Stopped Respecting Intent
This is where most people who are relatively new to Google Ads are often caught out.
Phrase match used to mean exactly what it said. If you bid on "teeth whitening" as a phrase, the search had to contain that phrase, but Google's keyword matching now works on meaning, so even an exact match brings in close variants that share intent rather than wording.
What I said on the audit call, and still believe, is this: "Google doesn't really understand the difference. It just thinks they're all the same."
It used to be a lot more controllable, but the platform now makes a judgement about what a searcher meant, which means two service lines that share a vocabulary look exactly the same to that judgement.
This means that the general dentistry campaign, which was bidding on broad practice terms, was picking up a long tail it never chose:
Around a third of that campaign's spend was going to searches with clear orthodontic intent. All of those clicks were paid for at general dental rates, which in Australia typically run between $8 and $15 for a term like "dentist near me," and every single one of them landed on a page that couldn’t actually help.
The click was never the waste. The mismatch was.
Problem Two: Budget Decides Which Campaign Wins, Not Relevance
This is the part that most business owners find genuinely surprising.
When two campaigns in the same account are eligible for the same search, they don’t bid against each other. Google resolves it internally and serves the one with the higher Ad Rank. There is also a separate Google policy that stops two of your ads running when they point at near-identical landing pages.
So there’s no bidding war, but a quiet, permanent selection process. Ad Rank is heavily influenced by the budget of a campaign.
The general dentistry campaign for our client had a budget around eight times higher than the orthodontics one. It won "orthodontist near me" nearly every time, not because it was the better answer, but because it was the bigger campaign.
This means that the orthodontic clinic was more or less being outbid by its own sister practice for its own core service.
Problem Three: It Leaks Both Ways
This is the detail an audit will miss if it only looks into the campaign that is obviously bleeding.
When we ran the search terms report on the small orthodontics campaign, we expected it to be clean, but it wasn’t. Around a fifth of its already low budget was going to general dentistry queries, arriving as exact-match close variants on terms like "dentist" and "dental clinic."
This means the small campaign, which was already starved, ended up spending part of what it had on searches that belonged to the campaign that was drowning it.
Neither campaign had a fence. The bigger one just won more often.
That reframes the fix: this is not about restraining a greedy campaign, but the account having no boundary between two services at all.
Problem Four: Everyone Was Judging It on a Number That Was Wrong
This goes back to those "six conversions."
Around half were "get directions" clicks. The client's reaction when we found it was correct: "Why would that even be set up as a conversion action?"
My instinct was that it was actively poisoning the bidding by teaching the algorithm to prioritise people who were already on their way in. They're people who are already on their way, which means you are not winning them, you are paying to count them.
There is a real version of that failure, because only primary conversion actions feed Smart Bidding, and stacking low-value actions as primary genuinely does water the signal down.
When we checked afterwards, directions were already correctly set to secondary, so the bidding wasn’t being corrupted.
I was wrong about the mechanism, but the problem was still real because the reported conversion count was what everyone in that room, agency and client, had been using to judge the campaign for months. It said six, but the number of people who actually asked to become a patient was three.
Something worth adding is that a directions click is a genuinely poor signal for a clinic because some of them are things like staff searching the practice name on their commute to work.
The Real Blocker Was Not Technical
All of the fixes for the issues above take a couple of hours of work. The thing that had stalled the campaigns for months wasn’t technical at all.
Both clinics had separate budgets and separate people accountable for them. The general dentistry budget holder didn’t want their money funding a page that also promoted orthodontics, while the orthodontic side did not want to carry general dental traffic.
This resulted in the account doing the one thing nobody had chosen: spending the larger budget on searches it could not serve, all because no one owned the decision to stop it.
To their credit, the client said it before we had to: "I don't know if there's an elegant solution to that." There isn’t. There’s only the decision about who owns the searches, and then funding that decision.
That is an organisational problem wearing a Google Ads costume. If you are running more than one service line under one roof, this is the version of the problem you are most likely to actually have.
What We Did, In Order
We deliberately sequenced so the cheap, reversible work happens before anything that needs a budget conversation.
The order is more important than it seems. Steps 1 to 3 cost nothing and don’t need approval from anyone. They also produce the clean data that makes step 5 an evidence-based conversation instead of a turf war.
Check This In Your Own Account
Narrow checks, specific to this problem. For a broader sweep, our service business Google Ads self-audit covers the general health of an account, and our dental practice lead generation guide goes deeper into this sector specifically.
If the first check doesn’t turn anything up, the rest are probably fine. If it turns up several, the others will almost certainly confirm it.
What We'd Do Differently
Two things.
We would have run the search terms report on the small campaign first, not the big one. The obvious problem was the campaign that was wasting money. The insight was in the one that wasn’t because it proved the leak ran both ways and that the account had no intent boundary whatsoever. We found the right answer, just later than we needed to.
We would have stopped trusting the conversion count earlier. Everyone in that room was experienced, the reported number still shaped how a five-figure monthly spend was judged for months, but nobody opened it up to see what was going on inside. A conversion count you haven’t personally taken apart is a rumour, not a metric.
The wider lesson has nothing to do with dentistry. If your business is selling two things that share a vocabulary and both live in one ad account, you should assume they are contaminating each other until you have looked closer. Google isn’t going to work out the difference for you.
The good news is that the fix is mostly free, and the first part of it is a search terms report you can run immediately.
This article is based on a real client audit. The industry, business details and figures have been changed to protect client confidentiality, but the pattern, diagnosis and fix are all exactly as they happened.
If you have more than one service line running through a single Google Ads account, and you have never checked whether they are competing, this is a worthwhile exercise.
We are happy to run the search terms report with you and tell you plainly whether there is a boundary problem or not, with no pitch or obligation attached if the account is clean.
Book a time here: https://calendly.com/leadtreemarketing/30min
