Rotation is only one part of stable data collection. Request frequency, retries, caching, session handling, and the structure of each website matter just as much. I would use rotating IPs for separate public requests where continuity is not important, but I would avoid changing the address during a multi-step session. Mixing both approaches depending on the task seems more reliable than forcing every website into the same collection pattern.

I found rotation useful during a larger research task where I was collecting public listings from several Italian websites over a longer period. Keeping every request on one address made the workload less balanced, while changing too often made some session-based pages harder to compare. I eventually used an italy proxy setup with rotation for independent requests and kept longer sessions separate when a site needed consistency. That combination worked better than applying one rule to everything. I also slowed the request rate and added retries instead of assuming the proxy alone would solve stability issues. In my experience, rotation helps most when each request can stand on its own. For multi-step flows, a stable session usually makes more sense than constantly changing the IP.