I would also separate the initial trial budget from the amount approved for routine runs. That makes it easier to spot a test that accidentally became a permanent process. Review the first batch of output before raising its traffic limit. The page at simplynode.io/locations/new-zealand-proxies has plenty of useful information on New Zealand proxies, billing terms, and session modes. Read those terms alongside your own usage logs rather than treating a low unit price as the whole budget. Track the cost of each usable batch, including retries that produced nothing. That gives the team a practical number to compare when choosing between two collection methods.