Doubling your budget doesn't double your result
If you double investment in a campaign, the likelihood that the result also doubles is small. I give this example in the class: a campaign at €1,000 per month moved to €2,000, a 100% increase, but revenue might rise only 70% or 60%.
I want to understand whether the additional revenue justifies the additional spend after accounting for the relevant costs. Returns can diminish as the budget grows, and that behaviour can change over time. I therefore compare the results again rather than assuming last month’s spending level remains appropriate today.
The sweet spot and campaign fatigue
I call this optimal point the "candy point" or the "Nutella point", the investment level where I extract the maximum return before the curve starts to decline. This point is not static, and the older or more fatigued a campaign gets, the lower it drops.
Fatigue is closely tied to how many creative variables are running at once. If I launch few variations of creative, copy or hook, that point drops drastically in a short time. If I work with multiple variables simultaneously, testing different angles, pain points and benefits, the decline is much slower across months. That's why, when managing large budgets, you need a constant flow of new creatives, new copy and new angles, otherwise exposure frequency spikes and cost per result climbs out of control.
Growing an online business: what I learned about margin, team, and AI
What I expect from a media buyer in the AI era
The real example of the book campaign
I share in the class the experience I had promoting my own book, where cutting the budget kept the same sales volume. At the start, daily spend ranged between €40 and €100, with good results. Over time, the audience became saturated and results held steady even at €10 per day.
Over time, most of the immediate audience had already bought the book and the audience became saturated, responding less and less to the same creatives. At that stage, I gradually reduced the budget down to €10 per day, and sales stayed practically the same as with €20 or €30. For that specific campaign, the sweet spot had already dropped to €10, and doubling or tripling the budget did not generate the same returns. Spending an extra €20 to gain 10% more sales made no sense at all.
This is a historical example, from one specific campaign and one specific product, not a rule that automatically applies to every business. But it illustrates the reasoning well: sometimes cutting the budget while keeping the same absolute sales volume is the decision that protects margin.
Testing sensitivity to ROAS and profit
There is no way to know if you've found your optimal point without testing budgets above and below the current one, raising and lowering values over time while measuring actual profitability. That's how sensitivity testing works, comparing the same asset's results across different periods.
As I say in the class, you raise and lower budgets over time and measure your profitability along the way, that's how you do sensitivity testing. This means comparing a given budget at one point in time with other budgets on that same asset, over time, to understand whether raising it paid off or whether cutting it in half actually improved returns. This reasoning applies differently to small budgets: if a campaign generates no return at €1 per day, it's unlikely to generate return at €30 or €100 per day either, because the root problem usually isn't solved simply by putting more money into it.
Looking at the business as a whole, not product by product
Not every product or campaign needs to be profitable in isolation, as long as the business, as a whole, makes money by month-end before other fixed costs. Some campaigns exist to capture customers who later buy other products, generating value over time.
There are entry-level products or campaigns with lower sales friction that serve to capture customers, even at lower direct profitability, because they later feed into a ladder of products and services that generates value over time, what I usually call customer Lifetime Value. Understanding this requires looking at attribution models: recognizing whether a campaign that doesn't sell much directly is, nonetheless, assisting sales elsewhere in the funnel. Cutting spend purely by looking at each line's isolated profitability in a spreadsheet, without looking at the business as a whole, is a mistake I see often among people managing large budgets who want fast results without that broader view.
Broad audiences instead of over-segmentation
With large budgets, going too narrow on audience targeting tends to backfire rather than help. I describe in the class what I call the "X effect": when segmentation is too fine, exposure frequency rises quickly, sales volume drops and cost per result climbs at the same time.
In that class, I recommended a broad targeting layer as the base of the strategy. I gave the concrete example of restricting only by age, from 20 to 50, with no other filters on interests, gender or geography, letting the algorithm work with the largest possible sample of people. More segmented strategies, like sequential remarketing to people who already saw certain ads, work as a complement to the main structure, not a replacement for it. Trying to scale a large budget inside very narrow audiences is one of the fastest ways to spike frequency without sales keeping pace.
A practical recommendation to check your saturation point
If you want to apply this reasoning to your own business, you could start by isolating one stable campaign, one that already has some maturity, and comparing two comparable time periods at clearly different budget levels, looking at actual profit rather than just ROAS.
The comparison criterion should be the profit that actually remains after subtracting product and fulfillment costs, not just the ROAS shown by the platform, which tends to inflate the campaign's apparent performance. This is a suggested exercise, a concrete way to put the sensitivity test described in the class into practice, not a fixed formula that replaces your judgment about your specific business. Every product, every margin and every audience has its own sweet spot, and you only find it by testing directly. The question is never spending more to sell more, it's understanding with discipline at what point that extra euro stops being worth it.
Source note
This article is based on the class I gave in Sextas Ímpares #91, "Como Gerir Grandes Orçamentos em Anúncios," available at youtube.com/watch?v=SrJOaLpJG28. It's worth watching from 00:23:00, where I talk about testing sensitivity to ROAS and profit, and from 00:26:30, where I share the book campaign example. To better understand the reasoning on audience breadth, listen also from 00:19:00.