Two inputs decide it
There are two things you actually need to know before you pick a number.
First: what's the one event that matters? For a subscription app, it's paid conversions, with trial starts as a proxy in the early weeks. For B2B, it's qualified opportunities, not raw form fills. For e-commerce, it's margin-positive purchases, not revenue.
Second: what does that event cost you to buy right now? If you don't know, that's the first thing the budget needs to answer.
Once you have those two inputs, you can back into a number instead of guessing.
Sizing to your stage
The platform needs roughly 50 conversions per ad set per week to optimize. That's the threshold. But you don't have to point optimization at your final conversion event from day one. You can point it at the richest signal your starting budget can generate.
For an app, that might be trial starts before you have enough paid conversions. For a B2B tool, it might be demo requests before you have enough closed deals. The event you optimize toward first should be close enough to revenue to predict it, and frequent enough for the platform to learn from.
Validate lean, then scale toward the event that pays. The classic mistake runs in both directions: funding a scale ambition with a validate budget, or spreading a validate budget like it's already time to scale.
What the numbers look like in practice
In accounts the team manages: roughly $2,000 to $3,000/month for B2B and e-commerce, $1,000 to $2,000/month for apps. That's the floor for building real conversion history before deciding whether to scale.
Those aren't magic numbers. They're the outputs of the same math applied to those specific funnels. Your number might be different. But the process is the same: work backward from your conversion event, its cost, and the volume you need to make the data mean something.