How the auction actually works

Meta does not sell you placements at a fixed price. Every time someone could see an ad, an auction runs, and the winner is not the highest bidder.

Meta ranks advertisers by your bid multiplied by how likely that person is to take the action you asked for, plus a quality adjustment. So an ad people respond to wins against a higher bid on an ad they ignore, and it wins at a lower price.

The practical consequence is the most important thing in this course. You cannot buy your way past a bad ad. A boring creative with a large budget is expensive and ineffective, and a genuinely interesting creative is cheap. Most of your improvement will come from the ad itself, not from targeting settings or bidding strategies.

Choose the right objective, because Meta optimises for exactly what you asked for. Ask for traffic and you get clicks from people who never buy. Ask for engagement and you get likes. Ask for leads or purchases and Meta finds people who do those things. The commonest expensive mistake here is running a traffic campaign and wondering why the clicks produced nothing.

The learning phase. A new ad set needs roughly fifty conversions to learn who responds. Below that, results are unstable and unreliable. This is why editing an ad set every day prevents it from ever working: each significant edit restarts the learning.

Budget properly. Very small budgets spread across many ad sets never exit learning, and it is better to run one ad set properly than five badly.