Hey, it's Ivi 👋
Like ever week, here's your Monday tip to start the week right.
The 48-hour rule.
You launch a campaign. Twelve hours in, you check the numbers. They look bad. CPM is high, no purchases, CTR is mediocre. You start tweaking. You bump the budget, switch the audience, pause the underperformer, change the headline.
Twelve hours later, the numbers still look bad. So you tweak again.
Giphy
Here's what's actually happening behind the scenes: Meta's algorithm needs around 48 hours (or roughly 50 conversion events) to exit what it calls the learning phase. That's the window where it's testing your ad across different slices of your audience to figure out who actually responds. During those first two days, the data is noisy and unreliable by design.
Every time you edit a campaign mid-learning, you reset that clock. The algorithm has to start over from zero. So the more you panic-tweak, the longer your ads stay in the noisy, expensive, unreliable phase, and the more budget you burn before you ever see a clean signal.
What counts as a "reset"? More than people realize:
Changing your daily budget by more than 20%
Switching or editing your audience
Pausing or duplicating an ad inside the active ad set
Editing the creative, headline, or primary text
Changing your optimization event or bid strategy
Basically, almost any meaningful change. The only edits that don't reset learning are minor things like fixing a typo in a description (and even then, Meta sometimes still resets it).
So the rule: don't touch a new campaign for the first 48 hours. No budget changes. No audience swaps. No pausing the "loser" ad. No new creatives. Let it learn.
There are real exceptions. Touch the campaign if:
Something is genuinely broken (dead link, pixel not firing, ad rejected)
Spend is wildly off (it's burning through 3x what you planned)
You're getting zero impressions after 6+ hours (a delivery issue, not a performance one)
Everything else, leave alone.
One more thing worth knowing. After the 48 hours is up and you've got cleaner data, don't judge a single ad in isolation. Look at the campaign-level cost per result first. A single ad can look like a loser inside a campaign that's actually profitable overall, and killing it too early can tank the whole thing.
This week's action: If you have a campaign that's less than 48 hours old, close the tab. Set a calendar reminder for the 48-hour mark and review it then. If you've got nothing running, save this email. Next time you launch something, this is the rule that saves you the most money.
See you next Monday, Ivi
P.S. Got a campaign you're itching to tweak right now? Hit reply and tell me how old it is. I'll tell you if you should leave it alone or actually fix something.
