How many ads should you actually be running?
A live calculator built on Andromeda-era data. Plug in your account - see the production target your campaigns actually need to compound.
Why volume isn't a vibe anymore
Imagine your ad account is a campfire. To keep it going, you have to throw new logs on. If you stop, the fire dies - fast.
Old Meta let you keep the same logs for weeks. New Meta (Andromeda) is a hungrier fire. The same logs only burn for two weeks. After that, the fire goes out, even if the logs looked great when you put them on.
The whole game now is: how many logs can you make, how different can you make them, and how fast can you throw them on?
In one line: there's a real number - for your spend and audience - of ads you should be running. If you're below it, you're losing money to fatigue. The calculator below tells you what it is.
live ads - top performers
ROAS lift at 20+ tests/mo
library refreshed per cycle
fatigue cycle at scale
Plug in your account.
Drag the sliders. The model pulls from public Andromeda-era performance research - fatigue cycles, similarity ceiling, top-quartile creative volume - so the recommendation reflects what actually compounds, not a vendor pitch.
ads running concurrently
cadence of replacement
× 2.5 variations each
net-new per refresh
The volume your account needs to keep Andromeda fed. Ship below this and the algorithm will keep replaying fatigued ads while ROAS bleeds out.
Sanity check: hand-cut economics
At $80–$300 per finished variation (editor time + footage + review), hand-producing 53 ads a month costs $4.2k–$16k before strategy or media. That's the production cliff most teams fall off.
Where the numbers come from.
Live ad count tracks public benchmarks: top-performing accounts run ~395 live ads, bottom-third ~296. Spend is the dominant input (with diminishing returns), audiences and campaigns multiply modestly.
Refresh cycle compresses to 2 weeks once spend exceeds ~$25k/mo because Andromeda exhausts creative faster at scale. Sub-scale accounts can stretch to 3 weeks.
Refresh share is the chunk of the library that needs replacing each cycle - ~28–34% based on observed week-on-week engagement decay (20–30% per week once fatigue starts).
Anti-patterns
The volume trap
Volume isn't an automatic win. Andromeda punishes four specific failure modes harder than under-shipping.
The production cliff is where AI-native tools earn their keep.
At a $25k/mo account, the calculator typically lands somewhere around 90–120 net-new ads per month. At hand-cut rates that's $7k–$36k of pure production cost - often more than the media spend it's supposed to feed. The teams compounding under Andromeda use AI variation engines to bring per-ad cost into single digits and free their senior creatives for concept work and final-pass taste, not assembly-line cuts.
Shuttergen is built for exactly this slot: take a winning concept, fan it out into 20+ on-brand variations with lineage tracking, ship them in fortnight cadence, and feed the model what it actually wants to see.
The playbook
Eight rules for sustainable volume
your team's coverage
Sources
What we read to build this
Stop guessing your production target.
Shuttergen turns one good concept into the 20+ ship-ready variations Andromeda actually rewards - at per-ad costs that make the math work.
Get started free