Two (weak) reasons I can think of: 1. If doing a low-volume of tracking requests, it could be an easier implementation instead of learning the ins and outs of an API. 2. No carrier approval required for web scraping, unlike the APIs.