[UPDATED MORE DATA] – Winning ad dies when moving it from Testing to Winning Ad Group

mjajou

Active Member

Winning ad dies every time I move it from Testing to BAU ad group — what am I doing wrong?​

App promotion campaign | Optimizing for in-app subscribes
I run a pretty simple campaign structure:
  • 1 Testing Ad Group — where I test new creatives
  • 1 BAU (Business As Usual) Ad Group — where I move winners to scale
I haven't found a winner in a while, so my BAU ad group currently only has 1 ad in it. This week I finally found one that looked legit, and then watched it die twice. Here's the full timeline:

Step 1: Found a winner in Testing​

After ~8,000 impressions in the Testing ad group, this ad was crushing it:

MetricValue
Spend$158.78
Results17 in-app subscribes
Cost/Subscribe$9.34
App Installs43
Cost/Install$3.69
Outbound CTR1.07%
Strong upper AND lower funnel metrics. I was hyped. Paused the ad and moved it into BAU.

Step 2: Duplicated via Post ID into a NEW BAU ad group — performance tanked​

I duplicated the ad using Post ID into a brand new BAU ad group (the ad lived in there alone). After ~2 days:

MetricValue
Spend$211.93
Results7 in-app subscribes
Cost/Subscribe$30.28
App Installs28
CPA went from $9.34 → $30.28 (3.2x worse). Spent more money and got way less results. Turned it off.

Step 3: Duplicated (Meta duplicate button) into an OLD paused BAU ad group — completely dead​

Thinking maybe the new ad group was the issue, I tried duplicating the original ad (this time using Meta's duplicate button, not post ID) into one of my older BAU ad groups that had been paused for a while.

MetricValue
Spend$71.53
Results0 subscribes
App Installs0
Outbound CTR0.13%

Absolutely dead. $71.53 spent with literally zero results. The 0.13% CTR vs 1.07% in testing tells me Meta was serving this to a completely different audience.

Summary of the decline:​

MetricTestingBAU #1 (Post ID → New Group)BAU #2 (Duplicate → Old Group)
Spend$158.78$211.93$71.53
Subscribes1770
Cost/Subscribe$9.34$30.28
Outbound CTR1.07%0.94%0.13%

What I've tried:​

  • Duplicating via Post ID into a new ad group
  • Duplicating via Meta's duplicate button into an existing (paused) ad group
  • Both methods killed the ad

Questions:​

  1. Is moving a winning ad out of Testing into BAU fundamentally flawed? Should I just leave winners in the Testing ad group and scale there?
  2. Does duplicating into a new or paused ad group reset Meta's learnings entirely, forcing it back into learning phase with no signal?
  3. Is the fact that the BAU ad group only has 1 ad in it hurting delivery? Does Meta need multiple ads to optimize properly?
  4. For those running a Testing → BAU structure: what's your process for graduating winners without killing them?
Any help is appreciated — I've now spent ~$440 trying to scale a $9.34 CPA ad and I'm getting nowhere. 😭
 
Last edited:
step 1 - the ad "was crushing" it with only $158 of spend sounds a bit overoptimistic. Was it absorbing all the spend of the testing ad group?
If yes, just keep it in there and convert that testing ad group into a new BAU so you can now have 2 assets spending and achieving your target cost but seeing the step #2, I'm pretty sure you are just seeing a false positive.

If after keeping it in the testing (now new BAU) ad group, it keeps spending and with good CPA (and better than the original winer) - just shift the budget there.

But again, I don't think that you found a real winner. Real winners absorb the spend wherever you put them very quickly.
 
step 1 - the ad "was crushing" it with only $158 of spend sounds a bit overoptimistic. Was it absorbing all the spend of the testing ad group?
If yes, just keep it in there and convert that testing ad group into a new BAU so you can now have 2 assets spending and achieving your target cost but seeing the step #2, I'm pretty sure you are just seeing a false positive.

If after keeping it in the testing (now new BAU) ad group, it keeps spending and with good CPA (and better than the original winer) - just shift the budget there.

But again, I don't think that you found a real winner. Real winners absorb the spend wherever you put them very quickly.
Got it, great points! Yes it was initially absorbing all of the spend in the testing ad group, before turning it off. But I agree, It does seem like a false positive
 
Back
Top