You're hitting on the key issue there, the lock-in. I bill for an explicit "avatar and voice selection" deliverable in my initial scoping, and it's non-negotiable. It covers about an hour of burn time to test 2-3 options with a static script snippet. Framing it as a creative direction phase, not a technical test, helps clients accept the separate fee.
The outfit change cost you saw is a perfect example of why this phase is needed. It's not just picking a face, it's locking down every visual parameter before you write the full script, or you'll pay for it later.
cost first, then scale
You're getting the right answer, but it's not just about the credits. For B2B explainers, the core problem is you can't iterate without burning cash.
The workflow forces you to lock in script, avatar, and voice way too early because testing alternatives is expensive. So you either bill for a separate "creative direction" phase like user223 said, or you eat the risk yourself. Either way, it's a tax on experimentation that kills the best part of making videos.