How I actually build a Google Ads account for a clinic

A single expert gives you a clean answer and a blind spot. The direct response person writes ad copy that makes a searcher feel understood, then forgets whether the account can even be measured. The technician builds a beautiful tight structure, then writes copy with no pulse. The strategist talks about moats while the actual ad group sits empty. Each one is right about their slice and wrong about everyone else's.

So I run all three over the same clinic before I build. I let them disagree out loud. Where they all land in the same place, I know that decision is safe. Where they pull apart, that's the actual decision I'm being paid to make. That's where the judgement lives.

Below are the three lenses in their own words, then how I reconcile them, then how I actually build and run the thing month after month.

Read these as three people sitting at the table, each certain they're the most important voice in the build. They're all partly right.

Starts from the searcher's emotional state the second they type. 'headache won't go away 11pm Sunday' is a different human to 'chiropractor near me 9am Monday'. The surgeon demands a scent trail with no breaks: the words in the query show up in the ad, and the ad's promise shows up on the landing page.

My physio background is the edge here. I know the daily tells of a sore knee or a stiff back, so I write to them until the reader thinks 'damn, these guys already understand my problem'. That's when marketing actually works, when it feels written specifically for the person searching it.

Believes structure compounds. Tight campaigns, tight ad groups, two ads running head to head in every ad group, disciplined negative keywords. Location based campaign almost always first, because no matter where you are in Australia, location based keywords are a strong campaign with high buyer intent and they show in the Maps pack. Then a condition based campaign.

The technician's hardest rule is the one I live by: it's not worthwhile running ads if you can't actually meaningfully measure them. Build it so the booking, not a page load, is what you count.

Looks above the keyword for the thing competitors can't copy. For me in 2026 that's two things. First, the measurement and data layer: clean conversion tracking plus de identified data from over 120 clinics telling each new account which headlines and keywords are already winning in similar clinics.

Second, the quiet truth that on Google you target the search intent, not the person. You can't buy a demographic the way you can on Meta. The moat isn't a clever keyword. It's the signal underneath it that nobody else has.

Put a real knee pain campaign in front of these three and watch them fight. The surgeon wants to write to the searcher who's scared they've torn something, panic language, late night searches, the works. The technician says fine, but split it into clean ad groups by body region so quality score doesn't tank and we can measure each one. The architect says hang on, what does 'knee pain' even mean to this clinic, because the data says two different clinics want two opposite humans behind that exact same search.

That last point is the one most agencies skip. The same keyword is a great client for one clinic and a terrible one for the next. The surgeon can write gorgeous copy for the wrong patient all day. The technician can measure the wrong patient to three decimal places. Only when all three are forced to agree on who we're actually chasing does the build hold together.

I don't try to resolve the argument by picking a favourite expert. I resolve it with a hierarchy of decisions that I run in the same order every time.

When the lenses pull apart, I run these decisions in order. Earlier ones outrank later ones. This is how three contradicting experts become one account.

And under all of it sits the only tie breaker that never loses an argument. It either makes sense or it doesn't financially. If leads come in around $110 to $120 and a client is worth $4,000 to $6,000 over their time with the clinic, that equation makes sense and we build. If it doesn't add up, no amount of clever copy or tight structure saves it.

Once the hierarchy has settled who we chase, the actual build is a bet I'm willing to write down. Pick a new client goal, estimate a realistic cost per acquisition for the area, multiply it out to a monthly budget, build to that, then check at month end whether the bet held. We make that hypothesis, build the ads around it, track it at the end of the month, ask were we right, and adjust from there.

I'm honest with owners about what that bet looks like in 2026. For generalised physio, $70 to $90 per new client is reasonable. I can't promise $50 consistently, and if you're paying $120 to $150 on a monthly average something is wrong. There's also a floor: below about $500 to $600 per campaign per month it's hard to get a return at all, which means your budget decides how many campaigns you run, not your ambition.

Compliance is the one line that overrides everything above it. I'll bid on 'counsellor' and 'therapy' for an adult mental health lane, but not on 'psychology', because Google makes you use the keyword word in your headlines and descriptions and AHPRA restricts psychology advertising tied to suicide prevention risk. A keyword opportunity that breaks the rule book isn't an opportunity.

Architecture first. The default is two to three campaigns split by axis, location based first because it's the safe, high intent bet that shows in the Maps pack, then a condition based campaign that's more of an assumption to be tested. I teach owners the structure in plain words: the campaign is the overarching thing and the settings, the ad group is who it gets sent to (the keywords), and the ads sit underneath. Each ad group runs two ads head to head.

I build around how people actually search, which is the injury or the problem, not the sport, the service line, or the job title. People are Googling the injury more than the sport. They search 'how do I fix this', not 'occupational therapist'. So the keywords are body regions and problems, and the ad copy speaks to the daily tells of that condition. On Google we target the search intent, whereas on Facebook we target the person. You can't target the person on Google, which is why elaborate demographic avatars matter far less here than owners assume.

Measurement is built in, not bolted on. I track the completed booking, not a page load, count One not Every so the same returning caller doesn't log twice, and keep the booking embedded or routed cleanly to the practice management portal so tracking sits near 100 percent instead of leaking out to 90. The landing page is bespoke, never a template, shipped as an HTML file you paste into a custom HTML block, and the whole asset (account, pages, tracking) is the clinic's IP with no lock in.

The build is the easy part. The cadence is the actual product the management fee buys, and it's adversarial by design. Google's job is to spend your money as fast as possible. Your job is to get as many new clients as fast as possible. Those are not the same goal, so you have to keep Google in line and say, that's not right, that's not right.

Every 72 hours I do negative keywording, because Google keeps widening your searches into weird, irrelevant territory to spend the budget. Every 7 days I review headlines, descriptions and keywords. Every 14 days I run a fresh champion and challenger split test: two ads compete, the loser gets cut, a new challenger is built to beat the winner. Sometimes one ad runs for a year because every split test I make loses to it. I like that. Data doesn't lie. You either win or you don't, and you decide off that.

Underneath that, every 7 days de identified data from over 120 clinics gets grouped by clinic type and matched to each clinic's ideal client, so a new physio inherits what's already winning for similar physios. Every clinic should get a better result for every clinic that joins. What I watch is simple: completed online bookings, phone calls longer than 90 seconds as a proxy for likely new patients, and whether the month end maths beat the hypothesis. The number coming in should always be bigger than the number you spent, otherwise you're borrowing from the future to pay for ads now. The monthly report lands by about the 5th, and it answers the three things an owner actually cares about: how many new clients, what we spent, what we paid per new client. Everything else is noise.