diff --git a/docs/my-website/docs/proxy/team_budgets.md b/docs/my-website/docs/proxy/team_budgets.md
index 854d6edf30..66ba679c65 100644
--- a/docs/my-website/docs/proxy/team_budgets.md
+++ b/docs/my-website/docs/proxy/team_budgets.md
@@ -4,6 +4,12 @@ import TabItem from '@theme/TabItem';
 
 # Setting Team Budgets
 
+
+# Pre-Requisites
+
+- You must set up a Postgres database (e.g. Supabase, Neon, etc.)
+- To enable team member rate limits, set the environment variable `EXPERIMENTAL_MULTI_INSTANCE_RATE_LIMITING=true` **before starting the proxy server**. Without this, team member rate limits will not be enforced.
+
 Track spend, set budgets for your Internal Team
 
 ## Setting Monthly Team Budgets
diff --git a/docs/my-website/docs/proxy/users.md b/docs/my-website/docs/proxy/users.md
index c812dccb19..d098e38de4 100644
--- a/docs/my-website/docs/proxy/users.md
+++ b/docs/my-website/docs/proxy/users.md
@@ -58,6 +58,9 @@ You can:
 
 **Step-by step tutorial on setting, resetting budgets on Teams here (API or using Admin UI)**
 
+> **Prerequisite:**
+> To enable team member rate limits, you must set the environment variable `EXPERIMENTAL_MULTI_INSTANCE_RATE_LIMITING=true` before starting the proxy server. Without this, team member rate limits will not be enforced.
+
 👉 [https://docs.litellm.ai/docs/proxy/team_budgets](https://docs.litellm.ai/docs/proxy/team_budgets)
 
 :::
@@ -793,6 +796,11 @@ Expected Response:
 
 Enable multi-instance rate limiting with the env var `EXPERIMENTAL_MULTI_INSTANCE_RATE_LIMITING="True"`
 
+**Important Notes:**
+- Setting `EXPERIMENTAL_MULTI_INSTANCE_RATE_LIMITING="True"` is required for team member rate limits to function, not just for multi-instance scenarios.
+- **Rate limits do not apply to proxy admin users.** 
+- When testing rate limits, use internal user roles (non-admin) to ensure limits are enforced as expected.
+
 Changes: 
 - This moves to using async_increment instead of async_set_cache when updating current requests/tokens. 
 - The in-memory cache is synced with redis every 0.01s, to avoid calling redis for every request.