Three fee models, and what each one is really pricing
Almost every quote you receive in Australia is one of three shapes, and the shape tells you what the agency believes the work is.
| Model | What it prices | Where it stops being fair |
|---|---|---|
| Flat monthly retainer | A defined scope of work each month, agreed before it starts | The scope grows and the fee does not move, so the work quietly shrinks to fit the fee |
| Percentage of ad spend | A share of what you pay Google, on the theory that more spend means more work | Spend and work stop moving together, and only one of them is being billed |
| Hybrid | A base fee for the standing work, plus a smaller share above an agreed spend level | Two moving parts, so both have to be defined or the invoice becomes unarguable |
A fourth model exists — payment tied to results — and it is rare for a reason worth understanding. It requires both sides to agree what a result is, to trust the same record of it, and to accept that whoever controls the measurement has a financial interest in the answer. Where the counting is genuinely independent, it can work well. Where it is not, you have swapped an argument about price for an argument about arithmetic, and that is the worse argument to be having.
None of the three is dishonest. What is dishonest is a fee whose model is never named, because then you cannot predict what happens when your spend doubles — or when the account needs nothing for a quarter.
What the fee is supposed to buy
Two quotes are only comparable once you know what sits inside each one. Most of the difference between a cheap proposal and an expensive one is not margin — it is scope that one of them has quietly left out.
Seven questions that turn a quote into a comparison
- Is the initial build charged separately, and what happens to that charge if we part ways in month two?
- Who writes the ads, and how often are new ones written rather than rotated?
- Who is responsible when the enquiry count is wrong — the agency, the web developer, or nobody?
- Are landing pages included, and if not, who is expected to build them?
- What happens when something breaks on a Saturday, and what is the actual response time?
- What is in the monthly report, and is it produced by a person who has looked at the account?
- Whose name is on the advertising account, and whose name is on the measurement setup?
The last question is the one that costs the most to get wrong and the least to ask. If the agency holds the account, everything built inside it — the history the platform learns from, the record of what has already been tried — leaves when they do. That is not a pricing detail; it is the difference between changing supplier and starting again.
Question three is the second-most expensive. Measurement sits between the advertising and the website, which is exactly where responsibility tends to fall through, and an account managed against a broken count is being managed against fiction. If nobody in the room will own the measurement layer, the fee is buying less than it appears to.
How to tell whether you are overpaying
Price on its own tells you nothing. A large fee against a genuinely active account can be excellent value; a small fee against a dormant one is money set on fire slowly. The test is the relationship between fee and work, and four signals say it has broken.
- Nobody can describe last month without a dashboard. A person doing the work can tell you what they changed and why in two sentences.
- The report is a list of activity. Impressions, clicks and tasks completed are inputs. If none connect to booked work, you are being shown effort rather than outcome.
- The recommendation is always to spend more. Sometimes that is the right advice. It should still arrive attached to the number that justifies it.
- The account has not meaningfully changed in six months. Stability is fine in a mature account. Stillness in an account that is not performing is a maintenance fee dressed as management.
A rough test, with placeholder figures
These numbers are invented to show the arithmetic; substitute your own. Say the fee is $1,500 a month and you ask, plainly, roughly how many hours go into the account. If the answer is around four, the implied rate is about $375 an hour; if it is around fifteen, it is about $100. Neither is automatically wrong — senior time on a complex account is worth more than junior time on a simple one — but the exercise moves the conversation from what feels expensive to what is being bought. The more revealing thing is whether anyone can answer at all.
A fee that scales with your spend is only fair while the work scales with it too.
The percentage model has a conflict built into it
If the fee is a share of spend, then the advice that increases the fee is the advice to spend more. That does not make percentage pricing wrong, and plenty of the businesses using it are honest. It means the recommendation to increase spend should never arrive on its own.
A fee that scales with your spend is only fair while the work scales with it too.
A defensible case for spending more contains two things: evidence that the cost of winning a job sits comfortably under what you will pay for one, and confirmation that you can service the extra work. Only the first is inside the advertising account; the second is yours to supply.
The flat retainer carries the mirror-image conflict, discussed far less. Once the fee is fixed, every additional hour reduces the margin on your account, so the incentive runs toward doing less. The protection is not a different model — it is a scope you can both point at.
Whichever model you choose, the fix for both conflicts is the same: agree in advance what the account is measured on, and hold to it. Once cost per booked job is the shared number, an argument about budget becomes an argument about evidence — a conversation that can be settled inside the account itself.
What the in-house alternative actually costs
Every agency fee is implicitly being compared with hiring someone. That comparison is usually made against the advertised salary, which is the wrong number.
The Fair Work Ombudsman's Pay and Conditions Tool gives you the base entitlement for the role you have in mind. On top of that sit superannuation, paid leave, payroll tax once your wages bill crosses your state's threshold, recruitment, software, and the productivity gap while somebody learns your business. Then the concentration risk nobody prices: one person means one opinion, one skill set and one holiday during which nothing moves.
None of that argues against hiring. It argues for comparing like with like. The useful version of the question is not "agency or employee" but "how many hours of senior attention does this account need, and what is the cheapest reliable way to buy them?" — which is a decision about breadth versus depth rather than about price.
One line item is routinely forgotten. If you are registered for GST, the fee that belongs in the comparison is the amount net of the credit you can claim, and the same applies to the ad spend underneath it. On a small budget that shifts the arithmetic enough to matter.
The price is the easy part of the contract. Ownership of the account is the part you will care about in a year.
The terms that matter more than the price
A year into an engagement, almost nobody is thinking about the monthly figure. They are thinking about how easy it is to leave, and what they take with them.
The ACCC publishes guidance on unfair contract terms in standard-form small business contracts, and the shapes it describes are worth recognising before you sign: automatic renewal with a short window to opt out, a clause letting one party vary the terms without agreement, and termination rights that only run one way. A fixed initial term is not itself a problem — there is real set-up work in the first months and it is reasonable to expect it to be paid for. What matters is that the term is stated plainly up front and that leaving does not cost you the asset.
The four lines to read twice
- Ownership. The advertising account, the measurement setup and the data are yours, named as such.
- Notice. How much, in writing, and whether it can be given at any point or only at renewal.
- Handover. What is transferred on exit, in what format, and within how many days.
- Variation. What can change without your agreement, and what cannot.
An engagement that expects to be judged writes these clearly, because they are cheap to honour when the work is good. Our ninety-day operating method and the terms we engage under both set out what stays with the business.
Common questions
Is a percentage of ad spend a rip-off?
No, but it needs a floor and a ceiling. It is a reasonable proxy for effort while the account is growing, and a poor one once it is stable. If the model is percentage-only, ask what happens to the fee in a quiet quarter and what happens if you double your budget in one month.
Should the set-up be charged separately?
Usually yes, because it is genuinely different work. The build is concentrated, one-off and heavy; the management afterwards is lighter and continuous. Bundling both into a flat monthly fee tends to mean either the first months are underpaid or the later ones are overpaid.
Who should own the Google Ads account?
You should. Google Ads distinguishes between administrative ownership of an account and access granted to it, and the difference only becomes visible on the day you change supplier. An agency should have access to your account rather than the other way round.
Is a lock-in period a red flag?
Not by itself. Real set-up work takes months to earn back, and a short initial term is a normal way to fund it. It becomes a flag when it renews automatically, when the notice window is narrow, or when leaving means losing the account history.