​
Understanding User Feedback

User feedback is crucial for improving the performance and accuracy of your LLM applications. It provides direct insights into user satisfaction and helps identify areas for improvement.

​
Types of Feedback

  • Binary feedback (thumbs up/down)
  • Numeric ratings (e.g., 1-5 stars)
  • Textual feedback

​
How to Pass User Feedback

When you are using Langtrace, you can pass both user id and user feedback using the Langtrace.withLangTraceRootSpan function for typescript or with_langtrace_root_span decorator in python. This helps with understanding usage patterns and measure the accuracy of your application from your user’s perspective.

​
Installation

Install and initialize Langtrace SDK. Refer to the installation guide for more information. 
// Must precede any llm module imports
import * as Langtrace from "@langtrase/typescript-sdk";
Langtrace.init({ api_key: "<LANGTRACE_API_KEY>" });

​
Implementing User Feedback

Use the Langtrace.withLangTraceRootSpan function to trace interaction which provides spanId and traceId for the interaction. 
import * as Langtrace from '@langtrace/typescript-sdk';
import OpenAI from 'openai';

// Initialize dotenv
import dotenv from 'dotenv'
dotenv.config()

// Initialize Langtrace SDK
Langtrace.init()

const openai = new OpenAI()

// Function to handle OpenAI interaction and user feedback
export const run = async (): Promise<void> => {
  await Langtrace.withLangTraceRootSpan(async (spanId, traceId) => {
    const response = await openai.chat.completions.create({
      model: 'gpt-4o-mini',
      messages: [
        { role: 'system', content: 'Talk like a pirate' },
        { role: 'user', content: 'Tell me a story in 3 sentences or less.' }
      ],
      stream: false
    })

    // Send traceId and spanId to your client
    // Collect user feedback (This is likely going to be another route in your application) For this example, we are sending the feedback immediately after the interaction
    const userScore = 5 // Example user score
    const userId = 'user123' // Example user ID
    await Langtrace.sendUserFeedback({ spanId, traceId, userScore, userId })
  })
}

​
Example user score

Collect the userScore and userId from your application and pass it to Langtrace.sendUserFeedback({userScore, userId, traceId, spanId})function along with traceId and spanId. 
import * as Langtrace from "@langtrase/typescript-sdk";
await Langtrace.sendUserFeedback({spanId, traceId, userScore, userId});

​
Best Practices

  1. Keep feedback collection simple and non-intrusive
  2. Clearly communicate how feedback will be used
  3. Act on feedback to improve your application
  4. Regularly analyze feedback trends

​
Privacy Considerations

When collecting user feedback, ensure you’re complying with relevant data protection regulations. Always inform users about data collection and obtain necessary consents.