Support here is not a tier of the product. Every account reaches engineers who operate the edge network rather than a script-reading layer in front of them, because most questions about a real-time dataset are questions about the network that produced it: why a record arrived when it did, why a region looks quiet, why two sources disagree about the same event.
That shape is possible because the answer is usually already in the data. Every record names the edge node that heard it, the point of presence that served it, and the provenance behind it, so a question about timing is a subtraction rather than an investigation. When you send us a record identifier, we are reading the same evidence you are.
What we ask in return is specificity. A stream name, a record identifier, and the window you were looking at turn a support thread into a measurement; "the data seemed slow yesterday" turns it into a conversation. The documentation covers the fields that carry the evidence, and the quickstart shows where to find them in a response.