utils/retention.py

def predict_retention(words):

    if not words:
        return 0

    duration = words[-1]["end"] - words[0]["start"]

    score = 100

    # too long → drop-off risk
    if duration > 30:
        score -= 30

    # too short → no engagement
    if duration < 6:
        score -= 20

    # weak opening signal
    if "..." in words[0]["text"]:
        score -= 10

    return max(0, score)