AI answers differ by location — here's how to see it
Ask ChatGPT “best family lawyer” from Denver and from Miami and you’ll get different answers — different names, different cited directories, sometimes different engines’ behaviors. AI answers are local whether you track it or not.
Why location changes the answer
Grounded answers are built on searches executed from the user’s region. Local intent queries pull local sources: directories, local news, regional review sites. Even non-local prompts drift by market — language, spelling, and which domains rank in that country’s Google all feed the fan-out.
The parameters
Both engines and the SERP accept explicit geo:
{ "prompt": "best family lawyer", "country": "US", "usState": "CO" }country covers 250 markets; usState drills to state level for the US. Google adds zip for pack-level granularity. One request, real localized result — no proxy fleet on your side.
Who this is for
- Local and regional brands: you’re not competing for “best X” nationally — you need to be the answer in your state.
- Multi-location businesses: one prompt set × your market list = a per-location visibility report.
- National brands: see where you’re strong and which regions name competitors instead.
- Agencies: per-market reporting is a client deliverable — see the local rank tracking use case.
How to run it
Build the matrix explicitly: prompt × market. Store market on every result row so reporting never mixes them. For the US, state-level sampling on your top prompts catches more than country-level on everything — a national average hides the states where you’re losing.
Full parameter reference in the API docs; pricing is per call regardless of geo, so a 50-state sweep costs the same per run as one.