Module 2: Foundations

Every account that performs sits on the same foundation: a structure simple enough for the algorithm to learn from, tracking that tells the truth, and a buyer who knows what each metric can and cannot prove. Build that foundation here, before you touch a live campaign in the next module.

Account structure: simple on purpose

The most common beginner account looks like a junk drawer: a dozen campaigns, each with a handful of ad sets, each spending a few dollars a day. It feels organized. It performs terribly. Modern delivery systems learn from conversion volume — they need a meaningful number of results flowing through each campaign to figure out who to show ads to next. Split $140 a day across fourteen campaigns and every one of them is guessing. Concentrate it into two and both of them learn.

The working rule is one job per campaign. Prospecting (finding new customers), retargeting (bringing back people who already know the brand), and retention or repeat offers are different jobs with different audiences, different creative, and different success numbers. Mixing them in one campaign makes the results unreadable: a blended cost per customer hides the fact that prospecting is failing while retargeting carries the account. Separate jobs, separate campaigns, numbers you can actually act on.

Inside that rule, keep everything else plain. Few campaigns, few ad sets, broad audiences, and your best handful of ads rather than thirty variants. Name things so a stranger could audit the account in five minutes: date, job, offer, angle. Structure is not where creativity lives in 2026. Creative lives in the ads. Structure's job is to stay out of the way, keep budgets concentrated enough to learn, and make every result traceable to a decision you made.

When you inherit a messy account — and you will — resist rebuilding everything on day one. Map what exists, note which campaigns carry the results, then consolidate in stages, watching the numbers after each change. A teardown that spends the client's money re-learning what the old structure already knew is not an improvement. Simple on purpose, changed on evidence: that is the whole doctrine, and Module 3 puts it to work on Meta.

You inherit an account with 12 campaigns, each spending about $8 a day, and none of them producing steady results. What's your first structural move?

Tracking set up right

Ads platforms optimize toward whatever you tell them a success is. If your tracking lies, the platform will diligently spend real money producing more of the wrong thing. So before any serious budget moves, three pieces get set up properly: the pixel, the server connection, and one conversion definition everyone agrees on.

The pixel is code on the client's site that reports what visitors do — viewed a product, added to cart, purchased. Browsers and privacy settings block a growing share of those signals, which is why the Conversions API exists: the client's server sends the same events directly to the platform, filling the gaps the browser missed. Run both together (the platforms deduplicate the overlap) and your data is as complete as 2026 allows. A buyer who checks tracking first will catch problems that look exactly like bad ads.

Then define the conversion — one event, named clearly, that represents money or the closest honest step to it. For a store, that's a completed purchase, not a page view. For lead generation, a qualified lead, not a button click. Every report, target, and optimization decision in the account keys off this event, so it gets agreed with the client in writing before launch, not argued about after a bad month.

Last, UTM discipline: every ad link carries consistent tags (source, medium, campaign) so the client's analytics can be lined up against the platform's numbers. The two will never match perfectly — different tools count differently — but with clean UTMs they should tell the same story. Inconsistent tags turn every monthly report into an argument. Complete the tracking setup sheet below for your practice account now, while the stakes are imaginary.

Your turn: Build the tracking setup sheet for your practice account (or a business you know well). Write down: (1) the ONE conversion event that counts as success and where it happens; (2) where the pixel is installed and how you verified it fires; (3) whether the Conversions API is connected; (4) your UTM template — the exact source/medium/campaign pattern every ad link will follow. If any piece is missing, write the step that fixes it.

A store's Ads Manager reports 300 conversions this week, but the store only received 40 orders. Before you touch a single ad, what's the most likely problem?

Reading the metrics without fooling yourself

The metrics stack reads like a funnel, and each level answers a different question. CPM (cost per 1,000 impressions) prices the auction: it tells you how expensive attention is in this market, not whether you're winning. CTR (click-through rate) shows whether the ad earns attention from the people who see it. CPC (cost per click) is where those two meet. CPA (cost per acquisition) and ROAS (revenue divided by spend) are the business numbers — the ones the client's margins judge. Problems get diagnosed by finding the level where the funnel breaks, not by staring at the bottom line.

Know what each number cannot prove. A high CTR cannot prove an ad makes money — curiosity clicks beautifully and buys nothing. A good ROAS inside the platform cannot prove profit if the tracking double-counts or the attribution window is generous. And no metric proves anything on a tiny sample: three sales on $60 of spend is weather, not climate. This is the discipline that separates buyers from gamblers — refusing to let a loud number outrun its evidence.

Attribution deserves its own honesty. Platforms count conversions that happened within a window after a click or view, and each platform claims credit generously. Line platform numbers up against the client's actual orders or leads, expect a gap, and decide with both in view. When a platform claims 100 sales and the store shows 70, the truth you report is the range and the likely reason — not the prettier number.

Your daily habit, then: read the stack top to bottom, find where it breaks, form one hypothesis, change one thing, and write down what you expected before you look again. Small samples get patience. Big claims get verification. In Module 3 you'll run this exact loop on live Meta structure — prospecting, retargeting, testing, scaling — and every decision there traces back to how honestly you read these numbers.

An ad has a 4% click-through rate — far above the account average — but after $200 of spend it has produced zero sales. The client wants to scale it because "people love it." What do you tell them?

What you can now do

Build accounts simple on purpose: one job per campaign — prospecting, retargeting — so each campaign gets the conversion volume it needs to learn, instead of a junk drawer of starving ad sets.
Set tracking up before any serious budget moves: the pixel plus the Conversions API filling its gaps, and exactly one conversion definition that everyone agrees counts as success.
Read the metrics as a funnel, each level answering its own question — CPM prices attention, CTR grades the ad, CPC connects them, CPA and ROAS judge the business — and diagnose at the level where it breaks.