Commission Based Pricing Month by Month: The Worked Example

Why a worked example, and not another argument

I have written twice before about why I think the hourly rate is a bad way to buy growth work, and why I do not sell consulting. Both pieces argue a position. This one does not argue anything. It takes one shop, gives it real numbers, and walks it through twelve months of the commission model that sits on my pricing page, month by month, invoice by invoice.

If you have opened the playground, dragged the sliders around and thought, fine, but what does that actually look like on a Tuesday in March, this is for you.

Two warnings before we start. The shop is invented, although every number in it comes from real engagements. And the whole thing is a worked example, not a promise. Whether your numbers move like this depends entirely on your shop, your margin, your tracking and how much of my attention the work genuinely needs. The exact path is agreed before we start, in writing, so there are no surprises. That sentence is on the pricing page, and it is in the contract, and I am going to repeat it here until it is boring.

The two models on the pricing page, in one paragraph each

There are two ways to buy an engagement from me, and people mix them up, so let me separate them first.

The phased packages. Data Science, Development, Digitalisation and Growth Plus. Each has a monthly price that falls in three phases: months one to four, months five to eight, and from month nine. The price follows my attention. The first four months are the build, when strategy, tracking, platform work and campaigns take most of my time. Then automation and dashboards take over, my hours drop, and so does the invoice. Those prices are final, published, and for business clients in the EU no VAT is added because my invoices fall under the reverse charge rules. The falling path is the typical case, not a rule: some projects need my full attention all year, and then the rate stays where it is.

The commission model. This one is only for ecommerce shops, and only where the shop already has enough revenue and clean enough tracking that my contribution can be measured honestly. It has a monthly base fee, and a commission calculated on the profit my work adds beyond the growth you would have had anyway. You pay whichever of the two is higher. Never both. If the commission beats the base fee, the base fee disappears. The mechanics live in Schedule A of the terms, the model lives in the playground, and this article is the bridge between the two.

Everything below is about the second model. I bring the first one back at the end, because the fair question is what the same year would have cost on a package.

Meet the shop

Let me invent a client. Call it a cookware and kitchen equipment shop near Linz. Family owned, twelve years old, a proper warehouse, a catalogue of about two thousand products from knives to professional ovens, selling to Austria and Germany with a little Switzerland. It does 1.8 million euros a year, which is 150,000 a month on average, with a spring bump and a big fourth quarter. Contribution margin after product cost, payment fees and shipping is 35 percent. Roughly 30 percent of customers come back within eighteen months, and a repeat customer is worth about one and a half times a first order over that period.

Marketing today is a Google Ads account run by a freelancer at 900 euros a month plus a percentage of spend, a Meta account that was switched off in 2024 because nobody could tell whether it worked, a newsletter that goes out when someone remembers, and analytics that were set up when the shop was built and have not been touched since. Ad spend is about 8,000 a month. The owner's honest answer to "which channel makes you money" is "Google, I think".

That last sentence is why the commission model exists. Not because the shop is broken. Because nobody can measure it, which means nobody can be accountable for it.

Month zero: the workshop and two weeks on me

Nothing is invoiced in month zero, and nothing is signed.

We start with the free workshop. Around a table, ideally in Linz, for most of a day. I look at the shop, the accounts, the data, and the team. The output is a written plan the owner keeps whatever they decide: what I would build, in which order, what it would cost on each model, and a forecast built with data science, best case, worst case and error margins, from their own history and industry data.

Then come two weeks of real work, on me, no strings. In this shop that means a tracking review, because the whole commission model stands or falls on measurement, and two or three quick wins: the search term report on the Google account has 40 percent of spend going to terms that have never converted, the checkout drops mobile users at the shipping step, and the newsletter has no automation at all.

At the end of the two weeks the owner has a plan, a cleaner ad account, and a view of how I actually work. They can walk away at that point, free of charge, with everything. Most do not, but the option is the point. If it fits, we sign.

The baseline, agreed before a single euro moves

This is the part people skip and it is the most important part.

Before the engagement starts we agree, in writing, a revenue baseline and a natural growth rate. They come from history, typically the preceding six to twelve months, adjusted for seasonality where the data supports it. For the cookware shop the twelve month history gives a baseline of 150,000 a month and a natural growth rate of about one percent a month, which is what the shop has been doing on its own with the freelancer and the newsletter and the seasonal rhythm. Seasonality is folded in as a monthly profile, so November is not compared against February.

Commission is payable only on measured performance above that baseline. Never on the baseline itself. If the shop does 150,000 in a month and would have done 150,000 anyway, I have added nothing and I am paid nothing beyond the base fee. That is not generosity, it is the model. The commission article on this site makes the point that agencies usually get paid for the growth the business was already making. Here it is written out of the contract.

The other things agreed at the same time: the tracking setup that will be the measurement (more on that below), the commission rate, the base fee, the minimum initial term, and the targets. The targets are planning tools. Missing them triggers no penalties on either side. Underperformance simply means little or no commission, which is the whole point.

The rule that does all the work

Every month I calculate two numbers.

The base fee, which in this example is 4,599 euros. It is agreed in writing, it is the floor of every commission model, and it cannot be waived, because senior work is delivered every month regardless of results. It keeps the lights on and commits me fully to the shop.

And the commission, which is a percentage of the profit my work has added above the baseline, adjusted for repeat customers. In this example the rate is 15 percent. The playground defaults to five; the rate is agreed per shop and depends on margin, on how much of the growth engine I am building versus running, and on how much risk I am carrying. Higher rate, lower base fee, or the other way round: the two move together in the conversation.

You pay whichever is higher. If the commission is 2,000, you pay 4,599. If the commission is 7,000, you pay 7,000 and the base fee disappears. Never both, never the base fee plus a commission on top. The terms put it as a formula: if the commission totals X and the base fee is Y, the total is Y plus (X minus Y), which is X. Written out like that it looks trivial. In a contract it is the difference between a model that aligns us and one that does not.

Now let me run the year.

Months one to four: the build

These are the intensive months, exactly as they would be on a phased package. In the cookware shop it looks like this.

Month one is tracking and foundations. Server side tracking so the numbers survive consent banners and ad blockers, a proper event taxonomy, consent mode configured correctly for Austria and Germany, UTM discipline across every channel, and a data warehouse that collects orders, sessions, ad spend and newsletter events in one place. The Google account gets restructured: brand separated from generic, the professional oven range gets its own campaigns because it has a completely different margin and buyer, and the search terms that never converted are excluded. Meta comes back on at a small budget purely to collect data.

Month two is the first real campaigns and the first automation. Abandoned checkout flow, a post purchase sequence that asks for a review at the right moment and offers the knife sharpening service to knife buyers, and a back in stock trigger. The mobile checkout shipping step is fixed. Product feed for Google Shopping rebuilt with proper GTINs and availability.

Months three and four are optimisation on data that now exists. The first dashboards go live in the owner's portal, per channel, per product group, refreshed on their own. Budget moves from the campaigns that do not pay to the ones that do, weekly. The first forecast is compared against the first actuals and the model is corrected.

Here is what the invoice column looks like. Revenue is what the tracking measured, baseline is where the shop would have been on its own, and the commission is what the formula produces, whether or not it is charged.

Month one: revenue 157,500, baseline 151,500, commission 457. Invoice: 4,599. Month two: revenue 165,375, baseline 153,015, commission 941. Invoice: 4,599. Month three: revenue 173,644, baseline 154,545, commission 1,454. Invoice: 4,599. Month four: revenue 182,326, baseline 156,091, commission 1,997. Invoice: 4,599.

The shop pays the base fee four times. The commission is nowhere near it. This is normal, and I say so in the workshop, because owners sometimes read "commission model" as "cheap at the start". It is not. It is a base fee at the start, and it becomes a commission model when the growth compounds.

What the shop has after four months: tracking it can trust for the first time in its existence, a Google account spending the same 8,000 with roughly a third less waste, Meta running with data behind it, three automations doing work nobody had time for, and dashboards the owner opens on their phone. And 32,000 euros of extra revenue above baseline, which at 35 percent is about 11,000 of contribution.

Months five to eight: optimise

My hours drop. Not to zero, nowhere near, but the daily work is now steering rather than building. Weekly budget allocation from the dashboards, creative testing on Meta with proper significance rather than gut feel, a second automation layer that segments the newsletter by what people actually bought, and the first serious use of the competitive data: the shop is losing the Dutch oven category on price to two German competitors by six percent and does not need to, because its delivery is faster and its reviews are better. We reposition instead of discounting.

Month five: revenue 191,442, baseline 157,652, commission 2,572. Invoice: 4,599. Month six: revenue 201,014, baseline 159,228, commission 3,181. Invoice: 4,599. Month seven: revenue 211,065, baseline 160,821, commission 3,825. Invoice: 4,599. Month eight: revenue 221,618, baseline 162,429, commission 4,506. Invoice: 4,599.

Eight invoices at the base fee. Look at month eight though. The commission is 93 euros short of the base fee. The curve is doing what compounding does.

Notice also what happened to the gap between revenue and baseline. In month one it was 6,000. In month eight it is 59,000. The baseline keeps growing at its natural one percent, the shop is growing at five, and the difference between those two curves is the only thing I am ever paid on.

Month nine: the commission overtakes

Month nine: revenue 232,699, baseline 164,053, commission 5,226. Invoice: 5,226.

The base fee disappears. Not reduced, not offset, gone. The shop pays the commission, because it is higher, and only the commission. From here on my income is a function of the shop's growth and nothing else.

Month ten: revenue 244,334, baseline 165,694, commission 5,986. Invoice: 5,986. Month eleven: revenue 256,551, baseline 167,351, commission 6,790. Invoice: 6,790. Month twelve: revenue 269,378, baseline 169,024, commission 7,639. Invoice: 7,639.

By month twelve the shop is doing 269,000 a month against a baseline of 169,000. That is 59 percent above where it would have been. It has been more than 25 percent above baseline for well over three consecutive months, which matters later, for the trailing commission.

The year in one table

Extra revenue above baseline across the twelve months: about 586,000 euros. Extra contribution at 35 percent: about 205,000 euros. Paid to me: eight months of base fee, 36,792, plus four months of commission, 25,641. Total 62,433. Kept by the shop: about 142,500 euros of contribution it would not otherwise have had. For every euro the shop paid me, it kept about 3.28.

Those are the same four numbers the playground shows at the bottom of its chart, computed the same way. And to say it once more: this is the target case, and the shop is invented. The next two sections are the ones that make the model honest.

What the commission is calculated on, exactly

Let me take month nine apart, because "commission on measured growth" hides four steps.

Step one, the growth above baseline. Revenue 232,699 minus baseline 164,053 is 68,646. Only this number is ever commissionable. The baseline is untouchable.

Step two, profit, not revenue. 68,646 at the 35 percent contribution margin agreed at the start is 24,026. Commission is calculated on profit. If a shop has a margin of zero, and some deliberately do in a land grab, revenue takes its place, but that is the exception and it is agreed in writing.

Step three, the repeat customer adjustment. New customers acquired this month will buy again. With a 30 percent repeat rate and a lifetime multiplier of one and a half, the factor is one plus 0.3 times 1.5, which is 1.45. So 24,026 becomes 34,838 of lifetime adjusted profit. This is the part that makes the model fair to me, and it is also the part the trailing commission clause exists to protect, because that lifetime value arrives after the month I earned it.

Step four, the rate. 15 percent of 34,838 is 5,226. That is the commission. It is higher than 4,599, so it is the invoice.

Every commission statement I send is itemised to this level and reproducible from the warehouse. The owner can review the calculation at any time. If we disagree, the agreed tracking data decides first, platform data second, and never my say so.

Where agreed, the commission can be calculated on a rolling three month average instead of a single month, to smooth seasonal swings. For a shop with a heavy fourth quarter I would normally recommend it, because otherwise December looks like genius and January looks like a collapse and neither is true.

How it is measured, and who can see it

The measurement is the tracking setup agreed at the start, not whatever the ad platforms claim. Google will happily tell you it caused a sale that Meta also claims and that the newsletter also claims. The warehouse has one order per order.

In practice that means server side tracking, consent mode done properly, UTMs on every link that leaves the house, a first party data warehouse, and attribution rules written down before we start rather than argued about afterwards. The shop keeps that setup, and my access to it, intact for the duration of the engagement and any trailing period. That is in the terms, and it cuts both ways: if the data is withheld or broken, billing falls back to estimates based on performance indicators, which neither of us wants, so nobody does it.

Who sees it: the owner, from day one, in the portal, on dashboards that refresh themselves. There is no monthly PDF. The commission statement is a view on the same data the owner has been looking at all month, so there is never a surprise on the invoice. If the owner's Steuerberater wants the calculation, it exports.

The slow case and the bad case

Now the year I do not put in the brochure.

Same shop, same everything, but the uplift I manage is two percent a month above natural growth instead of four. Perhaps the category is more saturated than we thought, perhaps a competitor goes to war on price, perhaps I was simply wrong about how much room there was. Month twelve revenue is 213,864 against a baseline of 169,024. Extra contribution over the year: about 95,000 euros. The commission peaks at 3,413 in month twelve and never overtakes the base fee. The shop pays twelve times 4,599, which is 55,188, and keeps about 40,000. For every euro paid, 1.72 kept.

That is still a year that paid for itself, and it is a year in which the shop ended with tracking, automation and a competitive position it did not have before. But it is not the target case, and a shop that expected month nine to arrive should know in the workshop that it might not.

And the bad case: uplift of half a percent, the extra contribution is a few thousand a month, and the shop pays 55,188 for less than it gets back. With those numbers the deal costs more than it adds, and the playground tells you so in red. So do I, before we ever sign. This is what the workshop forecast is for. If the best case does not clear the base fee comfortably, I do not offer the commission model, I suggest a smaller package or I tell you I am not your guy. The model is offered at my discretion, typically only where existing revenue justifies it, and that discretion is the reason it is fair rather than a gamble.

When targets are missed, and when you want out

Four situations, all in the terms, all worth knowing before rather than after.

Growth stalls. No commission accrues on the missing performance, only the base fee remains payable. Commission already earned in previous months is never repayable. No clawback, no set off, no reduction by later results. A good month stays a good month.

The shop causes the stall. Stock runs out on the best sellers, prices change against advice, tracking gets removed by a new developer, agreed campaigns are paused, the site goes offline for a relaunch. For that period the commission is calculated on the average of the three months before the event. This clause exists because I carry the downside of your growth, so I cannot also carry the downside of decisions I advised against.

You want to stop. Terms are monthly on the packages. The commission engagement carries a minimum initial term, typically six months, agreed in writing, because I am taking the risk on the front end. After that, notice as per the terms. Everything is documented so a successor can pick it up without me. Your accounts, your tracking and your data were yours from the start, and every piece of work became yours as it was paid for.

What happens to commission after you stop. No commission on new customers acquired after the termination date, and I remove the campaign tagging unless you ask me to leave it. But the customers I acquired during the engagement keep buying, and the lifetime adjustment in the commission was calculated on that. So a trailing commission runs for three to six months on repeat purchases from those customers, without the base fee. Where the engagement hit its milestones, by default revenue at least 25 percent above baseline for three consecutive months, the trailing period is twelve months. The cookware shop would be in that case. The clause is there to reward growth that was genuinely built and to stop an engagement being ended precisely to avoid paying for it, and it survives if the shop is sold or the sales are moved to a sister company.

What is not in the price

The fee is my work. It is not the things my work spends.

Ad spend is paid by the shop directly to Google, Meta and whoever else, from its own accounts, which it owns from day one. I never mark it up and I never hold it. In the example that is the 8,000 a month, rising as it proves it pays, and it is not in any of the numbers above.

AI usage is the shop's cost too. Where an engagement uses GrowCentric or other models for competitive analysis, content or forecasting, the token costs sit with the client. They are small next to ad spend, but I would rather say it than have it appear on an invoice unexplained.

Third party tools are the shop's subscriptions: email platform, review platform, the data warehouse hosting. Where a tool exists only because of my work, it is agreed before I add it.

And a data migration, if you are moving shops at the same time, is a one off next to the monthly fee, priced upfront on the pricing page by how much moves and where from.

The same year on the other models

The fair comparison, and the one I would want if I were the owner.

The phased packages. The cookware shop's engagement includes development on the shop, data science and performance marketing, which is the Growth Plus package. In euros that is 11,400 a month for months one to four, 9,700 for five to eight, and 8,000 from month nine. Over the same twelve months: 116,400. Digitalisation without the marketing would be 8,500, 7,200 and 6,000, so 86,800. On the commission model the shop paid 62,433 in the target case and 55,188 in the slow case. The package is more expensive and it is also more predictable, and it is what I recommend when the tracking is not yet clean enough to measure me honestly, or when the work is mostly platform rather than growth.

An agency retainer. A competent performance agency in Austria or Germany for a shop this size charges somewhere between 3,000 and 6,000 a month plus ten to fifteen percent of ad spend, so for the year roughly 50,000 to 85,000 before you count the fact that it does not build the tracking, the automation or the warehouse. Its income goes up when your ad spend goes up. That is the incentive I wrote about in the principal agent article, and it is the opposite of mine.

A hire. A senior growth marketer in Linz or Munich costs 70,000 to 90,000 fully loaded, plus tools, plus the developer they need for the tracking and the automation, plus the data person they need for the forecasting. If you can keep all three busy all year, hire them, genuinely. Most shops at 1.8 million cannot, which is the whole argument for fractional capacity.

None of these is wrong. The commission model is simply the only one of the four where my invoice goes down if your revenue does.

Why I price it this way

Principal agent theory says the person doing the work and the person paying for it want different things unless the contract makes them want the same thing. Hourly billing rewards hours. Ad spend percentages reward spend. Flat retainers reward not getting fired. All three are what I called, in the piece on the hourly rate, doing just enough to keep the client from walking away.

A base fee or a commission, whichever is higher, does something none of those do. For the first eight months of the cookware shop's year I earned the same whether the growth came or not, and I still built the whole engine, because month nine only arrives if it is built properly. From month nine my income is your growth, measured on data you can see, above a baseline you agreed, with no way for me to claim what you would have had anyway. The alignment is not a value. It is arithmetic.

And the part I care about most is not in the arithmetic. It is that the owner opens the dashboard on a Tuesday in March, sees where every euro of ad spend went and what it came back as, and has not needed to trust me for a single one of those numbers. That is what I am actually selling.