Understanding auctions: why you pay this price

Online advertising is a permanent auction market

Every time a page loads or a feed scrolls, an auction runs in a few milliseconds to decide which ad appears. You're not buying a fixed-price slot: you're competing with other advertisers for the same attention. Understanding this mechanism changes everything, because it explains why two entrepreneurs with the same budget get opposite results — and why "paying more" is almost never the right answer.

What really determines your cost

On Google as on Meta, the winner of the auction isn't the highest bidder, but the one whose bid × quality combination is strongest. The platform wants to show ads people find relevant, because that's what keeps them around. The result: a high-quality ad pays less for a better position than a mediocre ad that overbids.

:::key[Key takeaway] Lowering your acquisition cost more often comes through a better creative and a better page than through a higher bid. Quality is a discount on the price of attention. :::

The vocabulary that drives everything

Three acronyms appear everywhere, and confusing them is costly:

• CPC (cost per click): what you pay when someone clicks. Useful to compare ads, misleading as the only judge.
• CPM (cost per thousand impressions): what it costs to be seen a thousand times. It's the "price of attention" for an audience.
• CPA (cost per acquisition) and ROAS (return on ad spend): the only ones that speak business. CPA tells you how much a customer costs; ROAS tells you how many euros of revenue each euro spent brings in.

:::warning[Trap: optimizing CPC instead of CPA] A low CPC that never brings a sale costs more than a high CPC that converts. Always judge at the end of the chain (cost per customer), not at the click. :::

The learning phase: why patience is a tool

When you launch a campaign, the algorithm doesn't yet know who converts. It enters a learning phase: it tests profiles, observes results, then concentrates delivery on what works. During this period — often a few days and a minimum number of conversions — the numbers are unstable and misleading. Constantly editing the campaign sends it back to zero and prevents it from learning. The discipline is to set a test budget, let the algorithm stabilize, then decide on reliable data.

:::info[Concrete benchmark] Many platforms exit the learning phase around fifty conversions per week per ad set. Below that, expect "jumpy" numbers: it's normal, not a sign of failure. :::

Tools to understand and estimate before paying

Before spending, you can already scout the terrain. Several free or built-in tools give benchmarks:

• Google Keyword Planner: search volumes and estimated bid ranges for Google Ads, free with an account.
• Meta Ads Library: see, for free, every active ad of a competitor — a treasure for understanding which hooks are running.
• TikTok Creative Center: trends in formats, sounds and high-performing ads, openly accessible.
• Audience estimators built into each campaign manager: audience size and estimated reach by budget.

:::tip[Tip: legally spy on the competition] Meta's Ads Library and TikTok's Creative Center are public and free. Before creating your ads, list what your competitors have been running for a long time: an ad that runs for months is almost always profitable. :::

Automatic or manual bidding: what to choose at the start

Platforms offer to manage the bid for you ("maximize conversions", "target ROAS") or to set it by hand. For an entrepreneur starting out with no conversion data yet, objective-driven automatic bidding is almost always the right choice: the algorithm has more signals than you. Manual bidding becomes useful later, when you know your costs and want to impose a ceiling. Starting in manual "to keep control" is a common mistake: you deprive the algorithm of what it does best.

:::example[Concrete case] A coach launches their first Meta campaign with €20/day on "maximize conversions". They resist the urge to tune everything by hand, let it run 5 days, and get their first CPA data. Only then do they decide to adjust — on facts, not on a day-1 hunch. :::

Key takeaways

:::key[The chapter's key points]

• The auction is won by bid × quality, not by budget alone.
• Steer on CPA and ROAS, never on CPC alone.
• Respect the learning phase: set a test budget and touch nothing until the data is stable.
• Use free tools (Keyword Planner, Meta Ads Library, TikTok Creative Center) to estimate and get inspired before paying.
• At the start, prefer objective-driven automatic bidding over manual bidding. :::