CPC, CPA, CPM, ROI, retargeting and practical ad testing without the myth that every euro spent automatically returns two.
Advertising buys attention, not demand
Paid ads can send people to a project quickly, which makes them useful for testing. But an ad does not repair a weak offer, unclear audience, poor landing page or bad unit economics. It simply makes those problems happen faster and at a measurable cost.
Know the basic numbers
CPM is the cost of one thousand impressions. CPC is the cost of a click. CPA is the cost of a defined action. CAC is the full cost of acquiring a customer. ROAS compares ad-attributed revenue with ad spend, while ROI looks more broadly at profit after costs.
These metrics answer different questions. A cheap click can still be useless if the visitors do not buy. High ROAS can still hide weak profit when margins, refunds, support or fulfilment are ignored.
The funnel determines what you can afford
Imagine that 1,000 clicks cost €500. If 5% register, 50 people sign up. If 10% of those become customers, you bought five customers for €100 each. Whether that is good depends on contribution margin, repeat purchases, churn and lifetime value.
This is why “we will spend €1 and get €2 back” is not a strategy. You need to know where the conversion assumptions come from and whether the customer economics remain positive after all costs.
Most failures are not caused by the ad platform
Common problems include targeting people with low intent, sending everyone to a generic homepage, asking for too much commitment too early, weak creative, slow pages, broken tracking, a mismatch between the ad promise and the landing page, or simply an offer people do not want.
Retargeting can help — but only after interest exists
People who visited a pricing page, started checkout or used a free tool are different from people who merely saw an ad. Retargeting can be efficient because it focuses on warmer audiences, but it cannot create unlimited demand from a tiny pool.
Use small tests to learn, not to prove yourself right
Start with one audience, one offer and a few genuinely different messages. Define the success metric before spending. Stop or change tests that fail the economics instead of increasing the budget in the hope that the algorithm will rescue them.
Early campaigns are often research: Which audience reacts? Which problem statement earns attention? Does anyone continue to the valuable action? These answers can be useful even when the campaign is not yet profitable.
Ads work best after the project already has a signal
Paid acquisition becomes much easier when you know who converts, what they value, how long they stay and how much margin they create. Advertising is then a scaling tool for a working system rather than a substitute for product-market fit.