Paid advertising is one loop, repeated: you tell the platform what you want, fund it, and read what came back. Everything in this product hangs off those three moves.
What you tell the platform to optimize for — who it shows your ad to and what counts as a result. Pick the thing you actually want, not the one that looks cheapest.
The objective decides which people the platform hunts for: tell it "clicks" and it finds habitual clickers; tell it "purchases" and it finds buyers. That is why optimizing for a cheap proxy (clicks) rarely produces the expensive real thing (sales).
The tradeoff: deeper objectives need more data. If your budget cannot buy enough purchases per week to exit learning, optimizing one step up the funnel (add-to-cart, lead) gives the platform more signal to learn from — then move down once volume allows.
What your ads cost in the period — the input, not the outcome. Judge it by what it bought (results, at what cost per result), not by its size.
Three habits make results readable. Windows, not days: ad performance swings daily for reasons that mean nothing (auction luck, day-of-week). Compare complete weeks or 30-day windows.
Your history, not tables: industry benchmarks bundle together companies with different products, margins and audiences. "Better than my last 30 days" is a claim about your business; "above the industry average" often is not.
One change at a time: when several things changed at once, none of them can be credited. The accounts that improve fastest are the ones where each change can be traced to its effect.
Three levers matter far more than the rest: which objective you optimize for, how concentrated your budget is (see How much budget do I need?), and how fresh and varied your creative stays (see Which ad formats work now).
In ads by dotstack the agent proposes; nothing changes on your account until you approve it in the Approval Center. Sign up to open the campaign builder whenever you're ready.