Azure OpenAI Responses Telemetry

Implementation evidence

Azure OpenAI Responses Telemetry: from boundary to verified row

Use Azure OpenAI Responses Telemetry at a controlled application boundary, keep the event contract small, and verify a known outcome before building aggregate views.

1
Choose the outcome
Azure deployment cost attribution
2
Define the contract
provider, deployment, model, region, feature, and status
3
Instrument the boundary
Keep the Azure deployment name separately from the returned model identifier.
4
Verify the evidence
Exercise a known fixture, then inspect llm_request_completed for one correctly typed terminal row.

Before you start

Prerequisites and boundaries

A server-side Azure OpenAI credential and TELEMETRY_API_KEY
A stable deployment, region, feature, and prompt-version contract
A versioned price lookup for the deployed model

Delivery setup

Install and initialize server-side

Import telemetry-sh in server-only code and initialize it once with process.env.TELEMETRY_API_KEY. Keep ingestion credentials out of browser bundles, client-visible environment variables, source control, logs, and exception messages.

azure-openai-install

npm installation

bash

npm install telemetry-sh

1Prepare one reusable server-side delivery client with bounded network behavior.
2Add the outcome event at the success, failure, retry, or timeout boundary.
3Send controlled fixtures and inspect the stored rows before enabling an alert.

Snippet

Start with one structured event

Add this shape where the workflow completes, fails, or retries. Then build the dashboard from real fields.

azure-openai

Azure OpenAI Responses Telemetry event

javascript

import OpenAI from "openai";
import telemetry from "telemetry-sh";

const azure = new OpenAI({
  apiKey: process.env.AZURE_OPENAI_API_KEY,
  baseURL: process.env.AZURE_OPENAI_BASE_URL,
});
const startedAt = performance.now();

try {
  const response = await azure.responses.create({
    model: process.env.AZURE_OPENAI_DEPLOYMENT,
    input: userInput,
  });

  await telemetry.log("llm_request_completed", {
    provider: "azure_openai",
    deployment: process.env.AZURE_OPENAI_DEPLOYMENT,
    model: response.model,
    region: process.env.AZURE_REGION,
    feature: "document_summary",
    status: "success",
    latency_ms: Math.round(performance.now() - startedAt),
    input_tokens: response.usage?.input_tokens,
    output_tokens: response.usage?.output_tokens,
    total_tokens: response.usage?.total_tokens,
    prompt_version: "summary-v3",
  });

  return response;
} catch (error) {
  await telemetry.log("llm_request_completed", {
    provider: "azure_openai",
    deployment: process.env.AZURE_OPENAI_DEPLOYMENT,
    feature: "document_summary",
    status: "failed",
    latency_ms: Math.round(performance.now() - startedAt),
    error_type: classifyProviderError(error),
  });
  throw error;
}

Event contract

provider, deployment, model, region, feature, and status

input_tokens, output_tokens, total_tokens, latency_ms, and estimated_cost_usd

response_id, prompt_version, error_type, and accepted outcome

Implementation checkpoints

Checkpoint 1

Keep the Azure deployment name separately from the returned model identifier.

Checkpoint 2

Read usage from the response and apply the price version associated with the deployment.

Checkpoint 3

Do not log prompts, outputs, API credentials, end-user identifiers, or raw provider error bodies without explicit approval.

Verification

Prove the event arrived

Run this after exercising known success and failure cases. Replace the fallback table name if your final event contract differs from the snippet.

azure-openai-verification

Azure OpenAI Responses Telemetry verification query

sql

SELECT *
FROM llm_request_completed
ORDER BY timestamp_utc DESC
LIMIT 20;

Confirm one terminal row per logical outcome, with the expected status, identifiers, units, and UTC time.

Inspect the inferred schema and verify that retries do not change field types or generate a new logical event ID.

Search the stored fields for credentials, raw payloads, prompts, private content, and unbounded error messages.

Exercise a provider timeout, ingestion rejection, and process shutdown before treating the dashboard as complete.

Implementation references

Review the event contract, data-safety guidance, and upstream primary documentation before enabling a new production path.

LLM cost SQL Accepted output per dollar LLM cost template Azure OpenAI Responses API Azure OpenAI Responses reference

Production boundary

Keep the outcome event small and recoverable

This pattern provides

A bounded, SQL-ready outcome beside the upstream workflow.
Stable fields for dashboards, alerts, and cross-event correlation.
A fixture-driven path for validating success, failure, retry, and timeout behavior.

This pattern does not provide

An OTLP exporter, automatic collection pipeline, or replacement for detailed traces and diagnostic logs.
Exactly-once delivery merely because the payload contains an event ID.
Permission to collect raw provider payloads, user content, credentials, or regulated data.

Delivery and idempotency Batching and shutdown Troubleshoot ingestion

Event schema starting points

Review the row grain, emit boundary, required types, privacy classes, example payload, and validation checklist before adapting a query or snippet to production.

AI operations10 fields