Define the research question
Decide whether you need regional availability, public pricing, product descriptions or publication trends. Document your sample and exclusions. A larger proxy pool does not make a biased sample representative.
Match the network to the question
Use Datacenter where normal technical access is sufficient. Consider Residential only for a confirmed residential context, or ISP for repeatable checks. A mobile network is relevant only if carrier identity is part of the study.
Preserve provenance and uncertainty
Record URL, timestamp, region parameters, source permissions and transformation rules. Separate observed facts from inferred categories. Retain failed and excluded observations so downstream reviewers can see selection effects.
Make the run repeatable
Set collection intervals and concurrency limits, minimize personal data and define retention. Compare like-for-like snapshots and review source terms when the use changes. Confirm traffic measurement and cost separately from research record counts.