Most teams say they have a creative strategy. But if you listen closely, what they really have is a creative production plan.
Over time, I’ve noticed three ways teams think about creatives. And which one they’re at usually shapes performance.
| Stage | What the team chases | Where I stand |
|---|---|---|
| 1. Counting creatives | Volume | Where I never wanted to sell my ad production service |
| 2. Chasing IPM | IPM and CPI | Where I used to be |
| 3. Emotional connection | Meaning over mechanics | Where I am striving to be |
IPM belongs to the middle stage. Each stage has its own section below.
Stage 1: counting creatives
This is the “how many can you produce for us?” stage. It’s an output mindset. Teams think in cost per creative, not cost per winner.
The problem is that quantity without insight just raises your creative waste.
I’ve seen thousands spent on hundreds of creatives that produced exactly one winner, sometimes none. This mindset is especially common in app success stories built on cheap TikToks in the style of Gen Z.
Cheap production is expensive.
| Batch | Videos made | Winners found | Cost per winner |
|---|---|---|---|
| AI videos | 500 | 1 | 500 × the cost of one AI video |
| DIY videos | 30 | 1 | 30 × the cost of one DIY video |
Cost per winner is what a batch cost divided by the winners it found. The two batches are the illustrative example from my Andromeda playbook, not measured results. By my arithmetic, at the same cost per video, the AI batch pays 500 ÷ 30 = 16.7 times more for its one winner, so it only comes out ahead when one AI video costs less than 30 ÷ 500 = 6% of a DIY video.
Stage 2: chasing IPM
This is a more articulate mindset.
Here, you optimize for performance: measuring installs per mille (IPM), cost per install (CPI), and iterating toward efficiency. Higher IPM and lower CPI help ROAS when revenue per user holds up. Neither metric tells you whether it does.
At this level, data begins to guide design. You learn which messages, tones, or visual cues perform. You trim the fat, improve the funnel, and find your base hits.
But this stage has a ceiling. It’s still about what works, not why it works.
Stage 3: emotional connection
This is where the good teams live. You stop chasing metrics and start chasing meaning.
People don’t install because of mechanics. They install because something clicks.
Because your ad describes the app and mirrors how they want to feel. That’s when you see strange data patterns. IPM and CPI stay the same, but suddenly 20% of installs become purchases instead of 5%. I’ve seen it already multiple times in my accounts.
The numbers in the chart are illustrative, based on a pattern I have seen in several accounts.
Same IPM, same CPI. The share of installs that turn into purchases is what moves. Cost per purchase is CPI divided by that share, so by my arithmetic, at the same CPI, four times the share means each purchase costs a quarter as much.
AppsFlyer measured a similar split at scale, on day 7 retention instead of purchases. Their 2025 creative optimization report covers 1.1 million video variations across 1,300 apps from Q1 2024 to Q1 2025. In UGC on ad networks, the report compares gameplay reviews with testimonials. The ad that installs best is not the ad that keeps people.
Source: AppsFlyer creative optimization report 2025, UGC on ad networks. Gameplay reviews brought more installs per 1,000 impressions, and testimonials kept a larger share of the users they brought at day 7.
It’s why you see freezing people, annoying owls, or trapped kings. These moments make people feel something.
Look at the top 3 gaming ads in AppMagic’s rankings for the 90 days before I wrote this in October 2025 (subscription apps are no different). The first asks your IQ, the others ask you to save bears and save the town.
That’s why car ads talk about you taking yourself on an adventure, not the car taking you there. Function is hygiene, but emotion is what moves people.
Emotion makes your creative sticky. It gives people an anchor, a reason to open the app and make their first purchase. And in a world where AI can generate a thousand ads a day, thinking about them becomes your unfair advantage.