Almost nobody in this industry publishes what they charge, and the reason is not modesty. A price discovered on a call is a price set by how much the buyer appears able to pay, and the discovery process costs that buyer several hours of meetings before they learn whether the conversation was ever worth having. So this piece does two things. It explains the four ways agencies charge and what each model rewards when nobody is watching, because the model shapes the advice you get more than any individual person does. Then it puts our own numbers on the page and works out where they win and where they lose, because a company that publishes its prices and hides the arithmetic has published nothing.
Four models, four sets of incentives
Every agency invoice you will ever receive is one of four shapes, or a blend of two of them. The shapes matter because each one pays for something slightly different, and what a model pays for is what it quietly encourages more of. None of them is corrupt. All of them have a failure mode that shows up on a bad month.
The monthly retainer
A fixed fee per month for a defined scope. It is the most common arrangement in marketing services and the easiest to budget against, because the number does not move when your spend or your revenue moves. What it rewards is retention: the agency is paid to still be there next month, which aligns well with the slow compounding work — technical SEO, entity consistency, lifecycle email, review velocity — that nobody would ever fund on a project basis.
Its failure mode is coasting. A retainer paid for outputs rather than outcomes will happily produce four blog posts and a report forever. The defence is not a different model, it is a scope you can audit: name the disciplines, name what changes weekly, and insist that the reporting shows booked work rather than activity. If your agency cannot tell you which of its deliverables it would cut first when time is short, it has no priorities, only a checklist.
The project fee
A one-time price for a bounded piece of work: a website, a tracking rebuild, a brand system, a migration. It is the right shape whenever the work genuinely ends, and it is the wrong shape for anything that decays. The failure mode is the handover cliff. A project priced to finish has no budget for the month after it finishes, which is exactly when a new site starts losing rankings it used to hold, or a freshly built tracking layer meets its first plugin update. If you buy a project, buy a maintenance arrangement with it or accept that you are buying a snapshot.
Percentage of media spend
The agency takes a percentage of what you spend on ads. This is the model most likely to work against you, and it is worth being precise about why, because the usual objection is too crude. It is not that percentage-of-spend agencies are greedy. It is that the model makes the agency's revenue a function of your budget rather than of your results, so every piece of advice it gives you has a second meaning.
Recommending more budget is also recommending a raise. Recommending that you pause a channel that is not working is volunteering for a pay cut. Recommending that you move spend from a platform that charges you media into a discipline that does not bill on media at all — organic search, email, review generation, conversion rate work — is recommending that the agency shrink. None of those recommendations become impossible under this model. They just become expensive to give, and over a long engagement, expensive advice gets given less often.
There is a legitimate case for it, and it is worth stating fairly. Above a certain media volume the management load really does scale with the spend: more campaigns, more geographies, more creative in rotation, more pacing, more approvals, more people who need reporting. An agency running six figures a month of media for a flat fee designed around a five-figure account will either lose money or quietly ration attention. What makes the model defensible is a taper — the percentage steps down as the budget climbs, because a bigger budget is more work but not proportionally more work. A flat percentage that never tapers is the version to walk away from. It is also the version that gets most expensive precisely when you are most successful.
Performance and commission
Payment per lead, per booked job, per sale, or a share of incremental revenue. It sounds like the model that finally aligns everyone, and it does align one thing very tightly: the agency will chase whatever the contract counts. That is the problem as much as the promise. Pure performance pricing selects for demand capture over demand creation, because capture converts sooner and is easier to attribute, so the work drifts toward branded search and remarketing — the traffic that was largely going to convert anyway — and away from anything that pays off in month nine.
It also makes measurement the contract. Once a number decides an invoice, every disagreement about attribution becomes a billing dispute, and attribution disputes are unwinnable by design: platform-reported conversions, modelled conversions and your own records will never agree exactly. If you go this way, define the countable event in the CRM rather than in an ad platform, agree in writing who arbitrates a discrepancy, and expect to spend real time on reconciliation. A base fee plus a share of a clearly defined outcome is the version of this model that tends to survive a full year.
Ask what the model pays for. Over a long enough engagement, that is what you will get more of, whatever anyone intended.
What the ranges genuinely look like
The word agency covers a freelancer with two clients and a holding company with two thousand staff, so a single market average would be a fiction dressed as a fact. The stable thing is not the average, it is the structure: what any given discipline costs to buy on its own, from someone competent, on a continuing basis. Below are the bands we publish on our pricing page for exactly that — one discipline, bought standalone, per month. They are our own read of the market, they are stated as ranges rather than false precision, and they are the anchors the rest of this article does arithmetic against.
| Discipline | Bought on its own, per month |
|---|---|
| Paid media | $2,500 to $4,500 |
| SEO | $2,500 to $5,000 |
| AI search, AEO and GEO | $2,000 to $3,500 |
| Local and Maps | $1,500 to $2,500 |
| Analytics and tracking | $1,800 to $3,000 |
| Email and SMS | $1,800 to $3,000 |
| CRO and landing pages | $2,000 to $4,000 |
| Creative studio | $2,500 to $6,000 |
| Reputation | $1,200 to $1,800 |
| Web design and speed | $1,800 to $3,500 |
CurrentAds' own published anchors, not a third-party survey figure
Our numbers, in full
We sell ten disciplines and price them by how many of them are in scope, not by how much you spend on media. The ladder has four packaged tiers and one scoped one, and the published tier price is the price — there is no number between tiers and nothing below the entry rate.
| Package | Scope | Per month |
|---|---|---|
| Starter | One or two disciplines | $3,500 |
| Growth | Three or four disciplines | $7,500 |
| Scale | Five to seven disciplines | $12,500 |
| Dominate | Eight to all ten disciplines | $19,500 |
Above that sits a scoped tier for multi-location, franchise, enterprise and white label work, which starts at From $35,000 a month and is quoted rather than packaged, because the work multiplies per location and pretending otherwise is how an agency loses money on a rollout. There is also a one-time onboarding fee of $1,500, charged on day one because that is when the onboarding work happens: access, the tracking rebuild, the competitor teardown, the campaign architecture. It is waived entirely at $5,000 a month and above. The terms attached to all of it are on the guarantee page, including what is refundable and what is not.
The honest bundle arithmetic
Here is the part agencies do not publish. Take the ten standalone bands above, take the midpoint of each, and sort them cheapest first. Now ask what it would cost a buyer to assemble any given number of disciplines the cheapest way possible, and put that next to the package price that covers the same count. This is the least flattering version of the comparison we can construct: a real buyer choosing three disciplines might well pick three expensive ones, which would make the package look better. Sorting cheapest first assumes they picked the cheapest three.
| Disciplines | Cheapest assembly | Package | Package price | Difference |
|---|---|---|---|---|
| 1 | $1,500 | Starter | $3,500 | $2,000 more |
| 2 | $3,500 | Starter | $3,500 | identical |
| 3 | $5,900 | Growth | $7,500 | $1,600 more |
| 4 | $8,300 | Growth | $7,500 | $800 less |
| 5 | $10,950 | Scale | $12,500 | $1,550 more |
| 6 | $13,700 | Scale | $12,500 | $1,200 less |
| 7 | $16,700 | Scale | $12,500 | $4,200 less |
| 8 | $20,200 | Dominate | $19,500 | $700 less |
| 9 | $23,950 | Dominate | $19,500 | $4,450 less |
| 10 | $28,200 | Dominate | $19,500 | $8,700 less |
Midpoints of the published standalone bands, cheapest disciplines first
Where our own packages lose
Read the last column and the picture is not the one a pricing page usually paints. Starter never saves anybody money on the disciplines alone: at one discipline it costs more than buying that discipline outright, and at two it is exactly the same number. Growth costs more than the cheapest three-discipline assembly and only turns positive at four. Scale costs more than the cheapest five-discipline assembly and only turns positive at six. Dominate is the one tier that is cheaper than the cheapest assembly across its entire band, and by a widening margin as you add the ninth and tenth discipline.
So the sentence "our bundles are cheaper" is false as stated, and we are not going to write it. What is true is narrower and more useful: bundling gets cheaper the more of the stack you take, it is roughly break-even in the middle of each band, and at the bottom of a band you are paying for something other than a discount. That something is real — one team accountable for the whole surface, one measurement layer that every channel reports into, and no vendor able to blame another vendor — but it is coordination, not a price cut, and the two should be argued separately.
Two things cut the other way and we should say them too. The comparison above counts the ten disciplines only. It leaves out the things that ride on every tier whatever you pick — a senior operator rather than a junior pass-through, a live reporting portal instead of a monthly PDF, and a written weekly summary — which have their own published anchors on the pricing page and would move every row in our favour if they were added. And it uses midpoints; a buyer who happens to find the bottom of every range will beat the packages further, and one who lands at the top of every range will not. We have shown the version that is hardest on us because it is the version a sceptical buyer would construct anyway.
How to compare two proposals honestly
Most proposal comparisons fail because the two documents are not describing the same thing. The fix is mechanical. Convert both to a twelve-month all-in number and a scope list, then compare those two objects and nothing else.
- 01 Twelve months of management fee, plus any setup or onboarding fee, plus any minimum term you cannot exit.
- 02 Any percentage of media, priced at the spend you realistically expect in month nine, not month one.
- 03 Tooling. Call tracking, a landing page builder, a rank tracker, an email platform, a reporting layer. Whoever pays for it, it is part of the cost.
- 04 Creative production. If ad creative is billed per asset or excluded entirely, price the volume you actually need to test at.
- 05 Everything excluded that you still need, bought elsewhere at the standalone rate.
- 06 The exit cost. Offboarding fees, data export fees, and whether the ad accounts, pixels, audiences and analytics properties are in your name or theirs.
That last line is the one that most often changes an answer, and it is not really a pricing question. An account you do not own is a cost you cannot see until the day you leave, at which point it is the whole history of your advertising. We wrote out the specific clauses to look for in agency contract red flags, and the reasoning behind putting any of this in public in why we publish our pricing.
What price cannot tell you
A published number settles one argument and leaves the important one open. Price tells you what an engagement costs. It does not tell you whether the person on your account has run an account like yours, whether the measurement underneath it is sound enough for the numbers to mean anything, or whether the scope is the right scope. A cheap engagement pointed at the wrong discipline is more expensive than a dear one pointed at the right one, and both are worthless if the conversion tracking is wrong, which is more common than the industry likes to admit — the failure modes are set out in the broken tracking epidemic.
One disclosure belongs here rather than in the footnotes, because it bears directly on how much weight to put on anything above. CurrentAds has no paying clients yet. The single case study published on this site is an engagement with a company under common ownership with CurrentAds, published anonymised for exactly that reason; no revenue increase is attributable to the work, and none is claimed. Everything on this page is a commercial term you can hold us to, not evidence that the commercial terms produced a result for somebody else. If that makes the price list easier to judge and the pitch harder to believe, that is the correct trade.
One case this article does not cover well is the business whose numbers are real but whose budget is not there yet, because every model above assumes you can carry a full retainer from month one. That case has its own terms, including the reduced base and the single performance component, and they are set out on the startup and small business program rather than negotiated in private, for the same reason as everything else here.