$100 is enough to launch a small Facebook ads test, but rarely enough for Meta to optimise towards installs or purchases. The system needs about 50 optimisation events in a week to leave the learning phase.
Why the number of results matters
- Meta labels an ad set Learning limited when it is unlikely to get around 50 optimisation events in the week after its last significant edit.
- Delivery is less stable during learning, so cost per result moves more.
- Significant edits, such as changing the audience, creative or optimisation event, restart learning.
Worked example
If installs cost $2, then 50 installs cost $100, which is your whole budget for one ad set in one week. If your optimisation event is a purchase at $20, 50 purchases need $1,000 a week. These figures are arithmetic, not benchmarks; use your own costs.
In practice
With $100, test one ad set, one audience and a few creatives, and optimise for the cheapest event that still predicts value, such as installs or registration, rather than purchases. Use the test to learn which creative works, not to judge return on ad spend. Once you know your cost per event, budget for about 50 a week.
See how we budget Meta for apps on our Facebook ads agency page.
Sources checked 4 October 2026
Related Meta (Facebook) questions
- Are Facebook ads worth it?
- What's a good budget for Facebook ads?
- How much does Facebook pay per 1,000 views?
- How many followers do you need on Facebook to get verified?