A dental PPC agency can only see half your funnel, and it is not the half that pays
A dental PPC agency reports clicks, leads and cost. Your cases live in the practice records. What to settle about accounts, conversions and reporting first.
The short version
- What an ad platform counts as a conversion is a form submission or a phone call of a certain length. It is never a seated consult and never a started case.
- Settle ownership before month one. Your own Google Ads account ID, your conversion actions, your Meta dataset, your analytics property. Conversion actions created at an agency's manager account leave when the agency does.
- Keep one row per enquiry with the agency's source and cost fields beside your own outcome fields. If a started case cannot be walked back to a campaign, the reporting is decoration.
- At full-arch volumes a calendar month holds too few outcomes to read. Judge cost per seated consult over a quarter, grouped by the week first contact happened.
- Speed to first contact, contact rate and how many enquiries have any outcome written against them are the numbers with enough volume to review monthly.
Your agency’s reporting ends at the lead. Your practice’s records begin at the lead. Between those two facts sits the only question worth asking about paid media, and plenty of engagements run a year without anyone able to answer it: which campaigns produced the cases?
Nothing in Google Ads knows whether a case started. Nothing in your practice software knows what the click cost. Neither side is withholding anything; the join just does not exist. Until someone builds it, every monthly call is two people comparing numbers that were never about the same thing.
The account stops at the form
A conversion in a dental ads account is almost always a submitted form or a call past some length. Both are real events. Neither is a case.
So the cost per conversion in the platform is a cost per contact record. It will look good. It can be driven lower almost on demand by widening the audience, and the account will keep reporting improvement the whole time the surgical calendar stays empty.
None of which is anyone’s fault. The report that decides whether the engagement continues is produced by the party being judged, out of a system the other party cannot see into, measuring the only thing that system can see.
Settle the accounts before month one
This is boring, it is the part people skip, and it is worth more than any argument about campaign structure.
The Google Ads account should exist under your own account ID, with the agency attached as a manager. Built the other way round, the campaigns and the history are the agency’s property, and having paid for them does not change that.
Then a more specific question: where do the conversion actions live? One created at a manager account and shared downward disappears when the manager link is removed, and your campaigns’ conversion history goes with it. A year of learning, gone on the day you change agency. Ask the same about the Meta dataset and ad account, which belong in your own Business Manager with the agency granted partner access, and the analytics property.
None of this is suspicion. At $20,000 to $45,000 a case, an account that has been learning for a year is worth more than the retainer that built it, and anyone who has done this before has the answers ready.
One row per enquiry, with both halves on it
The fix is not a better dashboard. It is a single record, held by the practice, with both sides of the story on the same line.
The agency writes its half: when the enquiry arrived, the campaign, the ad group or creative, the search term where there is one, and the spend attributed to it. Your team writes the other: when someone first attempted contact, whether they reached the person, whether that person was a candidate, the consult date, whether they sat in the chair, whether treatment started, and the reason the record closed. It is the same discipline the rest of your pipeline needs, and most dental CRMs are not set up for it.
Here is the test. It takes ten minutes. Pick a case you started last month and walk it backwards: when did they first contact you, what were they responding to, what did that cost. If nobody in the building can do that for a case already in the chair, nothing in the monthly report is load-bearing.
Your account is too small to read monthly
Take a practice adding three full-arch cases a month from paid media, as a worked example. A month where it does two and a month where it does four look like a swing of fifty percent, and almost all of it is the calendar. One patient’s financing approval landing on the 2nd instead of the 29th moves the number.
Agencies get fired on that noise. They also get kept on it, which is worse: the month that happened to go well buys another two quarters of the same mistake.
Full-arch paid media does not generate enough outcome events per month for a month to work as a unit of measurement. The readable window is a quarter, grouped by the week first contact happened rather than the month the spend was invoiced, for the reasons set out in reading a Meta cohort. Agree that in writing before the first invoice, along with one more thing: the most recent three weeks appear in every report labelled incomplete. Those enquiries are still inside their decision window, and averaging them in will always make the newest spend look like the worst spend.
What there is enough of to read every month
Plenty, as long as you stop asking the outcome numbers to do a job they cannot do at this volume.
Median minutes from enquiry to first contact attempt. The share of enquiries reached at all. The share with any outcome written against them, the one that quietly invalidates everything else: if a third of last month’s enquiries have no disposition, your cost per case is fiction however long a window you read it over. Then the share of spend going to search terms that cannot become a full-arch case, and the consult show rate.
Those produce hundreds of events a quarter even in a small account, they move week to week, and they lead the numbers that do not. A practice whose front desk is qualifying rather than just booking will see it here months before it shows up in case count.
An agency that brings those to the monthly call is managing a system that ends in seated cases. One that brings cost per lead and a chart of impressions is managing the report.
What a good one asks you for
On a first call, listen for what they want to know before they quote.
Your average case value and the spread. Your close rate from seated consult to started treatment. How many consults the practice can physically seat in a week. Who, by name, calls a new enquiry, and how fast. Whether outcomes can be exported from wherever you keep them.
They need those to set a budget at all, because it comes backwards out of the case math rather than out of a benchmark. A firm that quotes a fee and a recommended spend without asking has not established what it is optimizing toward, and it will default to the only target visible from inside the account. Which is the lead. Which is the half that does not pay.
