-
Unpopular opinion: Your Meta/PMax campaigns aren’t failing because of targeting, but “creative velocity”
Posted by on December 9, 2025 at 11:36 amI've been reading through the recent threads here–specifically the hiring firm struggling with Meta and the insurance guy facing high search CPCs.
I was in the exact same spot about 6 months ago. I had unlimited budget on Search but hit a ceiling, and every time I tried to scale to Meta or PMax, I burned cash. I blamed the "audience signals" or thought the leads were junk.
The reality check? It wasn't the targeting. It was that I was feeding the algorithm 2-3 static images while my competitors were feeding it 20+ video variations a month.
Since the recent algo updates (Andromeda, etc.), creative *is* the targeting. But as a media buyer, I can't edit video to save my life, and hiring an agency for "testing" assets is too slow.
The Workflow Shift:
I stopped trying to make one "perfect" ad. Instead, I started using an AI video agent workflow to brute-force the testing phase.
-
Scripting: I take my best-performing Search queries and turn them into 15-second scripts (e.g., "Why X Insurance is a scam" vs "How to save on X").
-
**Generation:** I run these through a generative video tool that handles the stock visuals, voiceover, and pacing automatically.
-
**The Numbers:** Out of 10 AI-generated videos, 7 are mediocre, but 1 usually catches fire. That one winner drops my CPA by 40%.
It's not perfect–sometimes the AI motion is a bit floaty and I have to re-roll a scene–but it lets me test 10 angles in the time it used to take to brief a designer on one.
For those of you stuck on Search: How many new video creatives are you actually testing per week on social? I feel like volume is the only lever left.
-
-
1 Reply
-
This like back in the 2000s when it was all about the volume of content and blog posts. People don’t need another blog post like they don’t need another ad. If your competitor is uploading 20 ads a week then their ads just suck and don’t work. This should be about quality over quantity.
Creative as an input in the ad auction has always been the case the last 7+ years. All Andromeda did was take it up a level and force anyone not thinking about this to think about it. With that said, some brands will find interest targeting works over broad which might do better than lookalike targeting in another account. All this to say, just having a crap ton of creative is not going to solve people’s problems if the rest of the ad account is not set up for success.
-
Hey they’re, it sounds like you have no clue of what resonates with your audience so you’re spinning the wheel of AI creative until you stumble on a ‘Winner’ by chance.
Real pro tip: Research your audience. what they want, how they feel, what moves them. Build a plan on how you can move them towards action. Choose the right channels based on that. Craft your creatives to align your goal with customer expectations. You’ll get it right the first time everytime. Marketing is not gambling on creatives, it’s logical decision-making.
-
This is maybe a good idea starting out but you should absolutely be compiling audience lists to pipe back into the platform for targeting and expansion. Without that, one way or another you’re going to hit creative fatigue by saturating your market
-
I have a company using one video and version with slight variation (2 total).
Meta sales campaign.
Only adjust ad copy for sales and promotions.
Generates 20x+ consistently.
Targeting is your problem. You add ao much content, the algorithm doesn’t know what to do. Make more thoughtful, data driven, psychology strong pieces of content. You don’t need 20+ pet month. Most of my clients can’t handle that anyway (cost for time alone would 2-3x their budgets).
In contrast, I have a client who’s company fluctuates so much that one combination was a winner last month and sub 1x roas the next week. Then the following week, back to selling. If I shut it off, added new content, and saw sales return, I would be incorrect that the content made the difference.
Log in to reply.