From what I've tested, TikTok does seem to have some OCR capability for text overlays, but it's not as sophisticated as their speech recognition. The algorithm primarily focuses on: Audio/speech content, Hashtags and captions, User engagement signals, Watch time patterns
Text overlays might contribute slightly, but they're not a primary ranking factor. Best practice is to include key terms in your actual caption and voiceover rather than relying solely on text overlays. Has anyone else run tests on this?
