Rankings are local by default
For anything with commercial intent, search engines weight proximity heavily. The same query returns a materially different page one city over, which is why rank tracking without location control produces numbers that cannot be compared week to week.
Search engines are aggressive about automated queries
Rank checking at any volume looks exactly like scraping, and datacenter ranges are the first thing blocked. Residential addresses with rotation spread the queries across ordinary looking connections.
Check the market you sell to
If you rank in five countries, that is five different result pages and five different competitor sets. City targeting narrows it further for local search, where the pack results change street by street.
What to buy
Residential rotating, with country targeting for national tracking and city targeting for local. Metered bandwidth suits rank checks well, the requests are small and infrequent.
Common questions
Will a proxy show me my real ranking?
It shows you the ranking for the location and language you request from, which is closer to real than checking from your own machine, but no result is universal. Search results are personalised by location, history and device, so treat a proxied check as a consistent measurement you can compare over time rather than as the one true position.
Do I need a different address for every keyword?
No, but you do need enough addresses that no single one carries an unnatural volume of searches. Search engines are unusually good at spotting one address running query after query. Spreading the work is what keeps the results clean, not slowing it down.
Which country should I check from?
The one your customers are in, and each one separately if you sell in several. A ranking measured from the wrong country is not a worse version of the right answer, it is a different question. This matters most for anything local, where results change between cities inside the same country.