// Knowledge base
A budget is a switch, not an expense
By Mitchel van Luit
3 min read
Advertising budget does not behave like a normal expense item: below a certain threshold it yields virtually nothing, and above that threshold it starts to work. This article explains why that is and how to plan a budget that can grow along with it.
In short
✔ Algorithms need a minimum number of conversions to recognise patterns; below that, you keep paying for the learning phase.
✔ Starting small is sensible, but keeping the budget the same after the proof of concept is not.
✔ Cutting budgets while things are going well is the fastest way to lose what you have built.
✔ More budget solves nothing if the problem lies in the measurement, the destination or the offer.
In most organisations, advertising budget is treated as a standard cost item: the less the better, and if necessary we halve it. That does not work here, because budget behaves more like an on-off switch than a slider. Below a certain threshold almost nothing happens, and above it, it starts to work.
The reason lies in how the advertising platforms' algorithms learn.
Why a lower limit exists
An advertising system optimises by recognising patterns in who converts. For that, it needs conversions, and not a handful, but enough to distinguish coincidence from pattern.
If too few come in per week, the campaign stays in the learning phase. That is precisely the phase where the cost per result is highest. You are then mainly paying tuition fees, without ever reaching the part where the system does its work.
That is why we work with a lower limit of approximately 1,250 euros per month for the media budget. That figure is not about our fee; it is about the bare minimum an algorithm needs. Below that threshold you can advertise, but what you are mostly buying is uncertainty.
Starting small is sensible, staying small is not
We always advise starting small. First a proof of concept: one clear outcome, one channel, limited target audience. That limits the risk and provides the information upon which you build further.
But as soon as that proof of concept is in place, extra budget is needed to stabilise and grow the result. If that doesn't happen, you'll nose-dive during takeoff: the algorithms lose the plot without enough data and budget, and you lose precisely what you've just built up.
The tricky part is that this moment often coincides with the time when internal questions arise as to whether things could be scaled back a bit, since things are running smoothly anyway.
Why cutting back on a working campaign is counterproductive
With an ordinary expense item, the rule is: spend less, cost less. Not here.
If you lower the budget of a campaign that is performing well, fewer conversions will come in. The algorithm receives less signal, becomes more uncertain about who to target, and the cost per result increases. You will then have spent less and have a more expensive campaign.
If you still want to spend less, do so by stopping campaigns that aren't performing, rather than chipping away at everything. Shifting budget from what isn't working to what is costs nothing extra and is usually the biggest win to be had.
Plan the growth in advance
The practical solution is simple and is almost never done: agree before the start on what result will cause the budget to increase, and by how much.
For example: as soon as the cost per application is below an agreed amount and that remains stable for two weeks, the budget goes up by a fixed percentage. Then scaling up is no longer a matter of discussion, but of execution.
Without that agreement, the following will happen: the campaign works, someone has to ask for internal approval, that takes weeks, and in those weeks growth comes to a standstill. With a decision-making cycle of two to six months, those are costly weeks.
When budget is not the answer
There are three situations in which more budget only makes the problem worse.
- The measurement is incorrect. Then you are just optimising harder in the wrong direction. Measure first, then scale up.
- The destination is not converting. If your ads work but your landing page doesn't, you're paying for traffic you're giving away yourself. Conversion optimisation is cheaper than more budget then.
- The offer is not selling anywhere. Advertising magnifies what is there. If there is no evidence of demand, more budget only makes the loss bigger.
Check those three before you flip the switch. In all other cases: if it works and you don't put any more into it, you are consciously choosing not to let it grow.
marketing psychology
The summary at the top (the “in short” block) lowers the barrier to entry: anyone who sees the core message first is subsequently more likely to read the whole article.
// Frequently asked questions
Questions about this subject.
Why does a small advertising budget often not work?
Because the algorithm is not receiving enough conversions to recognise patterns. The campaign then gets stuck in the learning phase, which is precisely the phase where the cost per result is highest. You end up paying the tuition fee without getting the returns.
What is a realistic lower limit?
We allocate roughly 1,250 euros per month for media budget, with the willingness to scale that towards 10,000 euros if the campaigns are profitable. That lower limit is not about our costs, but about what an algorithm needs to learn.
Can we reduce the budget if things go well?
That's possible, but it is often the wrong time. With a lower budget fewer conversions come in and the algorithm loses signal, causing the cost per result to rise again. You then save money in the exact place where you are generating a return.
When is more budget not the answer?
If the tracking is wrong, if the landing page does not convert, or if the offer does not sell anywhere. In those cases, more budget only magnifies the problem. Fix that first, then scale up.
// Read more
More from the knowledge base.
Article: 5 min read
The Biggest Mistakes in Setting Up Tracking (and How to Avoid Them)
Tracking is one of the most important parts of your website [...]
Read on →
Article: 4 min read
How to get more reliable data with server-side tracking
The world of tracking is changing rapidly. Browsers are increasingly blocking [...]
Read on →
Article: 5 min read
Marketing integrations: which systems you connect and the benefits it brings
Many organizations invest in marketing, but forget that the maximum [...]
Read on →



