How to Tell if an Ad Campaign Actually Works

ROAS, CPA, attribution: the 4 numbers to check in 5 minutes to know if a campaign is making you money or just burning budget.

You spend on ads but you're not sure if it's working. You're in good company: 70% of small ecommerces don't know how to correctly compute their ad ROAS. Here are the 4 numbers that are enough.

1. ROAS — Return on Ad Spend

How much you sell for every euro spent on ads. ROAS 3 means for €1 spent you got €3 in sales. Below 2 you're almost certainly losing money once margin and costs are factored in.

2. CPA — Cost per Acquisition

How much you spend in ads to acquire a single customer. Always compare it to average margin and customer lifetime value (LTV).

3. CTR — Click-Through Rate

Percentage of people who, seeing the ad, click on it. High CTR means the message resonates. Low CTR = wrong creative or wrong targeting.

4. Paid-traffic conversion rate

Often half or a third of organic — and that's normal. What matters is whether it grows over time: if it drops, something in the funnel is broken.

The attribution problem

A customer sees your Instagram ad, then searches on Google, then buys. Who sold it? Meta will say 'me', Google will say 'me'. The truth is they both contributed. That's why you should always look at totals over time, not single touchpoints.

Never judge a campaign in the first 7 days: the algorithm needs to learn. Evaluate after at least 2 weeks and 50+ conversions.

Keep reading

Want to go deeper? These related guides build on what you just read: [how to measure marketing ROI](/guides/how-to-measure-marketing-roi), [where your customers come from](/guides/where-do-customers-come-from), [the simple marketing dashboard](/guides/simple-marketing-dashboard).

FAQ

What's a good ROAS?

Depends on margin. At 30% margin you need ROAS of at least 3.5 to not lose. With high margin (70%+) ROAS 2 can be enough.

Why does Meta show one ROAS and Analytics another?

Different attribution models. Meta counts every conversion that was viewed or clicked. Analytics follows last-click. Both are valid but measure different things.