The starting position
The pattern below comes from behavioral health programs with two or three locations. Typically the site is three years old, has a homepage and an about page and little else, the phone routes to a front desk that closes at five, and the previous agency reported keyword rankings.
Baseline numbers before any work started: starting monthly calls — pull from the account and verify with Tim calls a month, starting monthly ad spend — verify with Tim in ad spend, and no record connecting either one to admissions.
The client is confirm with Tim whether this engagement can be named.
Step one: make the phone measurable
Nothing gets spent until calls are attributed. Dynamic number insertion on the site, a distinct number on each Google Business Profile, and recording turned on where state law permits it.
The first week of recordings changes the plan more than any keyword research. It tells you what share of calls are patients and what share are vendors, how many ring out, and what the intake team says when someone asks about insurance.
In this case the recordings showed what the first week of call recordings revealed — verify with Tim.
Step two: build the page structure the site was missing
One page per level of care, per location. A page per payer relationship. A referrer page. Then internal links from each of them up to the location they belong to.
This is unglamorous work and it is the part that compounds. New pages create eligibility for queries the site could not previously appear for at all, and unlike ad spend they keep working after the invoice stops. It also removes the duplicate pages competing with each other, which is the most common structural problem we find on a first audit.
Expect three to six months before this shows in traffic. That lag is exactly why it goes second and not last: it has to start early to mature while paid search carries the census.
Step three: buy only the terms with intent
Exact match, tight geography, and a small keyword set: level of care plus city, insurance plus city, and brand. No broad match discovery until cost per admission is known and stable. Ad schedule set to the hours the phone gets answered, not the hours you post, with a separate after-hours plan rather than a blind bid.
For addiction treatment programs this waits on LegitScript certification, which gates the entire channel and takes weeks. Starting it late pushes the whole plan back.
Result over the first number of months in the reported window — confirm with Tim months: calls produced by paid search — verify with Tim calls at cost per qualified call — verify with Tim.
Step four: close the loop, then scale the cheaper channel
Calls get written into the CRM with their source attached, admissions status gets maintained by the intake team, and the outcomes get uploaded back to Google Ads as offline conversions with a value. Now the bidding chases admissions rather than form fills.
Only at this point is scaling a decision rather than a guess. Whichever channel shows the lower cost per admission gets the next dollar, and the honest answer is often that referrals and organic are cheaper but cannot absorb budget, so paid keeps the incremental spend.
What did not work is worth recording too: what we tried that failed in this engagement — get from Tim. Every engagement has one, and a strategy document that contains no failures has been sanded down for presentation.