1. Where the purchase happens. App Store or web. This one answer determines whether a percentage commission is even structurally possible, how you will be attributed, and how quickly you will see data.
2. What the attribution can and cannot show you. On Apple campaign links, a metric only appears once it reaches a minimum of 5 in the date range you are looking at, and Apple withholds or combines rows that represent very small groups of users. A campaign that sent three installs last week can legitimately look like zero. Knowing that in advance saves you from concluding a link is broken when it isn't.
3. Whether the terms are in writing. Rate, what counts as a qualifying result, when payouts run, and what happens to a refund or a cancelled trial. A developer-run program is not worse than a network one, but it does mean the terms live in an email rather than in a dashboard you can re-read, so get them written down.
4. Whether the program still exists. This is the one affiliate directories are worst at. Freedom, one of the most established names in blocking, currently states in its own help center that its affiliate program is paused while it changes. Notion, the generic productivity program most often recommended to creators, states on its own affiliate page that it is not accepting new affiliates. Both are still listed as open, with confidently wrong commission figures, across a dozen affiliate directories. Check the primary source before you write the post.
5. Whether the product survives a skeptical reader. Your audience will try to break it. If a blocker can be undone by deleting the app, toggling a setting, or waiting out a five second delay, the people who trusted your recommendation will find out within a week and the cost lands on you, not the developer.