429 Too Many Requests on bulk user updates — how to implement proper backoff

Trying to understand the precise retry logic for 429 responses when batch updating users via PUT /api/v2/users/{id}. My Newman run hits the rate limit immediately, and the Retry-After header is inconsistent. Should I parse the header or implement a fixed exponential backoff? Here is the current pre-request script failing:

pm.sendRequest({ ... }, (err, res) => { if (res.code === 429) { retry(res.headers['retry-after']); } });

Thanks.