Reading Data Day-by-Day
This is the #1 gap in every beginner's skill set. Setting up the campaign is easy. Reading what happens after you launch is what separates operators from people who just lose money. Here's exactly what normal looks like — day by day.
See the Day-by-Day table above for exactly what to expect after you launch. The output reading — how to interpret the data day by day — this generates more anxiety and bad decisions than any other single topic. Operators kill winning campaigns because day-1 data looked bad. They scale losers because day-2 ROAS was temporarily good. This module exists to fix that.
"High ROAS on day 1–2 means I should scale immediately." Wrong. Day-1 and day-2 ROAS is often inflated — Meta shows your ad to your best audience first, then expands. Premature scaling triggers a learning phase reset and frequently causes ROAS collapse. Wait until day 4–5 at minimum before making scale decisions.
"Low ROAS on day 1–2 means I should kill." Wrong. Early ROAS is almost always bad — the campaign is still in exploration mode. The #1 killing mistake in ecom is shutting down campaigns during the learning phase before the algorithm has found a delivery pattern. See the Day-by-Day table above.
"CPM tells me if my campaign is healthy." CPM is almost meaningless in isolation. High CPM with strong CVR = profitable campaign. Low CPM with no CVR = wasted spend. Read CPM only in context of ROAS, not as a standalone health signal.
"If ATC is high, the product will sell." High ATC with zero purchases almost always means checkout friction, not product-market fit confirmation. People who add to cart have interest — the problem is something between cart and payment page: surprise shipping cost, no PayPal, slow checkout. Fix that specifically.
Day 1 data almost always looks bad. That is completely normal. The #1 most expensive mistake beginners make is killing campaigns in the learning phase because day-1 numbers scared them. This module gives you the exact framework to tell the difference between "normal bad" and "actually dead."
What Normal Looks Like — Day by Day
| Day | Normal Pattern | What to Do |
|---|---|---|
| Day 1 | High CPMs, few/no sales, volatile ROAS. Meta is in early exploration. This is expected and not a signal to kill. | Do not touch anything. Do not reduce budget. Do not edit creatives. Let it run. |
| Day 2 | CPMs usually stabilize slightly. You may see first ATCs. ROAS is still unreliable. Do not extrapolate. | Check CTR and hook rate. If CTR is below 0.5%, your hook has a problem. Otherwise — do not touch. |
| Day 3 | Delivery becomes more consistent. First meaningful ATC data. Still within learning phase. Resist the urge to kill. | If you've spent 2× your target CPA with zero ATCs: check page speed and checkout on mobile. Otherwise hold. |
| Day 4–5 | Algorithm has enough data to begin optimizing delivery. CPMs often drop slightly. First purchase signals appear. | First real evaluation point. Compare ATC rate and CTR to benchmarks. Apply kill/scale logic from Module 10. |
| Day 6–7 | Campaign exits learning phase (if 50+ purchase events occurred). Delivery is now more efficient. ROAS becomes reliable. | Full kill/scale evaluation. This is the moment your data actually means something statistically. |
"15+ orders yesterday. 1 order today. Same spend. Nothing changed." This is not a signal. One bad day is noise. Use the 7-day rolling average, not daily numbers, to make decisions about your campaign.
The Metrics That Actually Matter — In Priority Order
The Funnel Diagnostic — Where Did the Money Go?
When results are bad, the problem lives in exactly one of four places. Work backwards from purchase to ad:
Daily Decision Tree — What To Do Right Now
This is the #1 most-requested visual in the course. Copy this mental model and run through it every morning when you open your ads manager. It eliminates panic decisions by replacing emotion with a structured diagnostic.
Open this page every morning before reviewing your numbers. Run top-to-bottom. The tree's purpose is to prevent emotional decisions — it gives you a mechanical process for every data state. The most expensive mistake in this business is killing campaigns based on day-1 data while they're still in the learning phase. This tree catches that.
Common Questions Operators Get Stuck On — Answered Here
Questions about ABO vs CBO, when to create a new campaign vs adding new ads, and how to structure your kill/scale decisions are all covered in detail in Module 12: Kill or Scale → — that module is the canonical home for campaign architecture logic.
If your data on Day 3 shows one pattern but doesn't match any of the scenarios above — drop your numbers in the Discord. People who've been through the same read will tell you exactly what it means.
Quick reference — ad funnel metrics
What each metric means, a healthy range, and the warning sign to watch.
🧮 Use the Break-Even ROAS Calculator for this step →
