Advanced · Chapter 0 · Last updated 2026-10-02
Start here: what Advanced covers and who it's for
Who this level is for
People who make a living from this. More precisely, at least one of these is true for you:
- You run ads for clients, and you're accountable for the numbers.
- Your own business already advertises steadily, and when costs rise you need to know which stage is leaking.
- You've installed the gauges from Intermediate, but what you can see isn't enough. There's a whole stretch of conversions Meta can't see at all.
- You can get ads running, but they fall apart as soon as you add budget, you don't know how to test creative, and when results drop you can't say why.
Don't start until you've finished Intermediate. Every Advanced chapter builds on what Intermediate set up.
Chapter 1 uses the event_id from Intermediate Chapter 1, and Chapter 2 (Chinese edition only) uses Chapter 1's fbc freezing logic.
Every chapter from Chapter 4 on uses the three lines you worked out in Intermediate Chapter 5 (stop-loss, target, scale-up) as its yardstick.
If you skip around, you'll find halfway through that something is missing, go back to fill it in, and get the order wrong as well.
Four parts, eleven chapters
Getting ads out doesn't make you a media buyer. Advanced has four parts: part one lets you see, and the other three teach you to make decisions with the numbers you can now see.
| Part | Chapter | What it solves |
|---|---|---|
| One: Completing measurement | 1. Meta CAPI server-side events | The part the browser eats |
| 2. LINE LIFF attribution (Chinese edition only) | The part that only closes after the customer moves to LINE, a messaging app widely used in Taiwan | |
| 3. Ad review: common rejection reasons and compliant copy | Blocked by ad review, with no idea which layer to change | |
| Two: Finding leaks and scaling | 4. Leak map | Which stage after the ad the money leaks from |
| 5. Account structure and scaling | Why results collapse when you add budget, and how to choose a structure | |
| 6. A creative testing system | Keeping testing separate from scaling, and knowing why a winner won | |
| Three: E-commerce in practice | 7. E-commerce campaigns: catalogs, dynamic retargeting and value bidding | How to run ads for a shop with many products |
| 8. E-commerce economics: order value, bundles and repeat purchase | How much you can afford to lose on a first purchase, and how to raise average order value | |
| 9. Peak seasons: before, during and after | How to prepare for a big sales event and how to wrap it up | |
| Four: Verification and diagnosis | 10. Incrementality: how much the ads really added | What the report says vs what the ads actually brought in |
| 11. When results drop: a diagnosis sequence and the weekly report | Finding which layer the problem is in by checking in a fixed order, and writing it up as a weekly report |
Part one covers three things Meta can't see
| What Meta can't see | What it costs you | Chapter |
|---|---|---|
| Browser events lost to privacy features and blockers (industry figures often cite 30–40%; Chapter 1 shows you how to measure yours) | The auction algorithm sees fewer of your successes | Chapter 1 |
| The whole stretch where someone adds you on LINE before they buy | On paper it looks like "the ads did nothing at all" | Chapter 2 (Chinese edition only) |
| Being blocked by ad review without knowing which layer to change | The whole ad set stalls, and then you edit an ad that was working until it breaks | Chapter 3 |
The technical bar, up front
Chapters 1 and 2 are different from the first two levels: you will install things, run commands and write config files. If you don't want to touch any of that, you won't get through Chapters 1 and 2. Chapters 3 to 11 involve no code at all. Of those, Chapters 3 and 4 can be read on their own, Chapters 5 and 6 go together, and Chapter 11 ties together everything before it, so read it last.
| What you'll do | How hard it is |
|---|---|
| Install Node.js and run a one-line deploy command in the terminal | Just follow the steps; you don't need to understand it |
| Copy ready-made code into a file and change two values | Just follow the steps |
| Read error messages and know where to look in the logs | This is where the real bar of this level is |
| Create two channels in LINE Developers and flip a switch that can only be flipped by hand | Lots of steps, none of them hard, but miss one and the whole chain fails silently |
If you don't want to touch code at all, you can have someone else do Chapters 1 and 2, but you need to be able to read the acceptance criteria. Each chapter's "How to tell it worked" and its checklist are there for you to check the work against. Check against those two; don't just take someone's word that "it's installed."
How the three chapters in part one depend on each other
- Chapter 1: Meta CAPI. The foundation. It does two things: it switches how you deploy the landing page from drag-and-drop upload to the command line (drag-and-drop upload can't hold back-end code), and then it builds the first endpoint. All of Chapter 2 is built on top of it.
- Chapter 2: LINE LIFF attribution (Chinese edition only). Depends on Chapter 1. The longest chapter in the set, with the most moving parts; eight in ten failures come from a missed link in the chain or steps done in the wrong order. If you don't send people to LINE, you can skip the whole chapter.
- Chapter 3: Ad review. Fully independent. No code, and you can read it whenever you like. If a rejected ad has you stuck, you can jump straight to it now.
How to read parts two to four
- Chapter 4, the leak map, comes first. Before any later chapter has you make a decision, you need to be sure the money isn't leaking somewhere other than the ads.
- Chapters 5 and 6 go together: structure decides how the money is split, and creative decides whether it's money well spent.
- Chapters 7–9 are only for e-commerce shops selling physical or digital products. Lead-generation businesses (Morning Light Yoga, for example) can skip Chapters 7 and 9, and in Chapter 8 read the repeat-purchase sections.
- Chapter 10 needs the three lines from Intermediate Chapter 5, the attribution setting from Intermediate Chapter 6, the three-way reconciliation from Intermediate Chapter 7, and at least 8 weeks of data with no major changes to your structure.
- Chapter 11 is the closing chapter. It strings the previous ten chapters into one diagnosis sequence, so read it last.
Example shops
| Xiaoman Desserts (fictional) | Morning Light Yoga (fictional studio) | |
|---|---|---|
| Where it stands in Advanced | Moved to a shop platform, three product types, free shipping on orders of NT$1,500 (New Taiwan dollars) or more, a daily budget of NT$1,500 and three ad sets, has repeat customers (assumed numbers) | Instant form → LINE → attends class → monthly pass |
| How Advanced uses it | Chapter 1 recovers the events lost on the browser side; the scaling, creative and e-commerce examples in Chapters 5–9 | Chapter 2: its sales actually all close in LINE; the lead-generation back-end examples in Chapters 4 and 11 |
| Chapters that apply | 1, 3, 4–11 | 1–6, 8, 10, 11 |
Morning Light Yoga used an instant form in Basic. In Advanced it runs into a very typical situation: after the form brings in the lead, the actual sale is negotiated in LINE, and Meta can't see that part at all. Chapter 2 is what solves this.
Time needed
| Chapter | First time through |
|---|---|
| 1. Meta CAPI | About 2.5 hours |
| 2. LINE LIFF attribution (Chinese edition only) | A full working day, plus three days of observation |
| 3. Ad review | About 1.5 hours (not counting the wait for the actual review) |
| 4. Leak map | About half a day, including going through checkout yourself once; lead-generation businesses add two weeks of logging reply times |
| 5. Account structure and scaling | About 3 hours, then 20 minutes a week; the scaling itself is measured in weeks |
| 6. A creative testing system | About 3 hours, then a 30-minute review each week |
| 7. E-commerce campaigns | About 4 hours, plus two days waiting for the catalog match rate |
| 8. E-commerce economics | About 4 hours, plus an observation period |
| 9. Peak seasons | About 2 hours 20 minutes; preparation starts a few weeks before the event, then 10 minutes a day during it |
| 10. Incrementality | About 3.5 hours, plus about 3 weeks of testing and observation |
| 11. A diagnosis sequence and the weekly report | About 2.5 hours the first time, then 30–60 minutes a week |
Don't skip the "three days of observation" in Chapter 2. That chapter has a whole section on why your own successful test counts for nothing.
The ceiling of this level
To be clear up front, so you don't finish and feel cheated:
This level lets you see accurately and decide with evidence. It doesn't guarantee you'll sell well.
Whether anyone wants the product, whether the price is reasonable, whether the landing page actually persuades people: running ads doesn't reach these things. Chapter 4 teaches you to measure them and take the numbers into the conversation, but whether they can be fixed is another matter.
What you will know when you're done is where the money stops, and where the next dollar should go. That's worth more than anything, but it isn't results in itself.
Key takeaways
- Part one of Advanced fills in three things Meta can't see; the other three parts teach you to decide with numbers: finding leaks, scaling, testing creative, e-commerce, incrementality and diagnosis.
- Only Chapters 1 and 2 involve code; Chapters 3 to 11 don't, and reading Chapter 4, the leak map, first is recommended.
- This level lets you see accurately and decide with evidence. It doesn't guarantee you'll sell well.
Chapter checklist
- I've finished Intermediate and have the three lines I worked out in Intermediate Chapter 5
- I know Chapter 2 (Chinese edition only) needs a phone that has never added my LINE Official Account
- I know which chapters I'll read (lead-generation businesses can skip Chapters 7 and 9)
Next chapter
→ Chapter 1: Meta CAPI server-side events
If a rejected ad has you stuck and you need it solved now → Chapter 3: Ad review. If you'd rather not touch code and want to deal with scaling and creative first → Chapter 4: Leak map.