Same reach, half the spend.

The two campaigns landed within 0.7% of each other on impressions — 16,463 against 16,571. That makes this an unusually clean read: the same amount of the market saw an ad in both cases, so what changed is what it cost to get there and what came back.

On that same reach, the Salubrum-run campaign spent $2,523 less and returned 194 more clicks.

Half the spend, the same reach, and roughly a third more clicks at a third more efficiency per impression.

The same slice of the Texas market saw an ad either way. One campaign paid $9.08 to bring someone to the site. The other paid $3.76.

The efficiency gap traces back to query discipline. Of the 100 most-shown search queries in the practice-run campaign, 19 were people looking for free care — free therapy, free counselling, free online chat — and another 20 were for services the practice doesn't offer. That is 39 queries in the top 100 that could never have produced a patient, costing $515 of the spend attributable to that sample.

In the Salubrum-run campaign, four. Nothing at all was spent on free-intent searches.

Tighter match types, a negative-keyword list maintained against the search-terms report each week, and campaign structure that put budget behind the highest-intent searches rather than averaging it across everything.

With the campaign running efficiently, a more fundamental problem became visible. A large share of the leads arriving — in both campaigns — were on insurance plans the practice cannot accept.

That is not something better keywords fix. Of the 100 most-shown queries in our campaign, seven mentioned insurance in any form, worth $6.98 of spend. Someone searching emdr therapy san antonio tells you their service and their city and nothing at all about their coverage. Google Search has no targeting dimension for insurance carrier — so the signal cannot be bought at the keyword layer, by us or by anyone.

Recognising that is what set the next phase of the engagement.

Paid social supports audience targeting built on third-party consumer data, which means payer likelihood can be filtered before the impression is served rather than discovered on a phone call.

That last condition is the point of the whole test — a direct attempt to solve, at the audience layer, the payer-mix problem Google Search structurally cannot address.

The test is ongoing and we are not reporting results from it yet. When it concludes we will publish the comparison, including if the audience data shows no lift.