Análisis del uso a lo largo del tiempo con la API de métricas de Copilot

Obtén información sobre cómo conectarse a la API, almacenar datos y analizar las tendencias de uso.

¿Quién puede utilizar esta característica?

Organization owners, enterprise owners, and billing managers

GitHub Copilot Business or GitHub Copilot Enterprise

En este artículo

Introduction

You can use the REST API endpoints for Copilot metrics to see trends in how users are adopting GitHub Copilot. During a rollout of GitHub Copilot, it's useful to view these trends to check that people are using their assigned licenses, see which features people are using, and understand the effect of your company's enablement plan on developers.

The API includes:

  • Data for the last 28 days
  • Numbers of active users and engaged users
  • Breakdowns by language and IDE
  • The option to view metrics for an enterprise, organization, or team

If you currently use the REST API endpoints for GitHub Copilot usage metrics, we recommend migrating to the REST API endpoints for Copilot metrics as soon as possible.

This guide demonstrates how to query the API, store data, and analyze a trend for changes to the number of users per week. The examples in this guide use the endpoint for an organization, but you can adapt the examples to meet your needs.

About endpoint availability

Endpoints are available to get data for an enterprise, organization, organization team, or enterprise team.

  • If you have a GitHub Copilot Business or GitHub Copilot Enterprise subscription as part of a regular organization or enterprise, you can use the endpoints for an enterprise, an organization, or an organization team. You don't have access to enterprise teams unless you're enrolled in a preview.
  • If you use a dedicated enterprise for GitHub Copilot Business—an enterprise account without the ability to create organizations—you can use the endpoints for an enterprise or an enterprise team.

Prerequisites

1. Create a personal access token

For our example, to get metrics for an organization, we'll create a personal access token (classic) with the manage_billing:copilot scope. See Managing your personal access tokens.

If you're using another endpoint, you may need different scopes. See REST API endpoints for Copilot metrics.

2. Connect to the API

We will call the API from a script and save the response as a variable. We can then store the data externally and analyze trends in the data.

The following example uses the Octokit client for JavaScript. You can use other methods to call the API, such as cURL or the GitHub CLI.

Example

In this example:

  • Replace YOUR_TOKEN with your personal access token.
  • Replace YOUR_ORG with your organization name, such as octo-org.
JavaScript
import { Octokit } from "octokit";

Import Octokit

const octokit = new Octokit({
  auth: 'YOUR_TOKEN'
});
const org = 'YOUR_ORG';

Set your token and organization

/*
const team = 'YOUR_TEAM';
const enterprise = 'YOUR_ENTERPRISE';
const entTeam = 'YOUR_ENTERPRISE_TEAM';
*/

Set other variables if required for the endpoint you're using

async function orgMetrics() {
  const resp = await octokit.request(`GET /orgs/${org}/copilot/metrics`, {
    org: 'ORG',
    headers: {
      'X-GitHub-Api-Version': '2022-11-28'
    }
  });
  const copilotUsage = resp.data;
  console.log(copilotUsage);
  }

Call the API

orgMetrics();

Call the function

// Import Octokit
import { Octokit } from "octokit";

// Set your token and organization
const octokit = new Octokit({
  auth: 'YOUR_TOKEN'
});
const org = 'YOUR_ORG';

// Set other variables if required for the endpoint you're using
/*
const team = 'YOUR_TEAM';
const enterprise = 'YOUR_ENTERPRISE';
const entTeam = 'YOUR_ENTERPRISE_TEAM';
*/

// Call the API
async function orgMetrics() {
  const resp = await octokit.request(`GET /orgs/${org}/copilot/metrics`, {
    org: 'ORG',
    headers: {
      'X-GitHub-Api-Version': '2022-11-28'
    }
  });

  const copilotUsage = resp.data;

  console.log(copilotUsage);
  }

// Call the function
orgMetrics();

Run the script locally

To test the script locally, save the file as copilot.mjs, then run node copilot.mjs.

Important

The .mjs file type is important. The import { Octokit } statement may not work with a regular .js file.

In your terminal, you should see output with a JSON array like the following.

[
  {
    date: '2024-11-07',
    copilot_ide_chat: { editors: [Array], total_engaged_users: 14 },
    total_active_users: 28,
    copilot_dotcom_chat: { models: [Array], total_engaged_users: 4 },
    total_engaged_users: 28,
    copilot_dotcom_pull_requests: { total_engaged_users: 0 },
    copilot_ide_code_completions: { editors: [Array], total_engaged_users: 22 }
  },
...

3. Store the data

To analyze trends over longer than 28 days, you will need to:

  • Call the API daily, using a cron job or scheduled GitHub Actions workflow.
  • Store data locally or with a database service such as MySQL.
  • Query the data to identify trends over time.

Example

In this example we'll save the data to a local .json file. To do this, we will import some modules for working with files, and update the orgMetrics function to save the response data.

The function saves new data that is returned each day, without overwriting old data in the file.

New steps are annotated in bold.

JavaScript
import { Octokit } from "octokit";

Import Octokit

import path from 'path';
import fs from 'fs';
import { fileURLToPath } from 'url';
import { dirname } from 'path';

Import modules for working with files

const __filename = fileURLToPath(import.meta.url);
const __dirname = dirname(__filename);

Declare variables for working with files

const octokit = new Octokit({
  auth: 'YOUR_TOKEN'
});
const org = 'YOUR_ORG';

Set your token and organization

async function orgMetrics() {
  const resp = await octokit.request(`GET /orgs/${org}/copilot/metrics`, {
    org: 'ORG',
    headers: {
      'X-GitHub-Api-Version': '2022-11-28'
    }
  });
  const copilotUsage = resp.data;

Call the API

  const dataFilePath = path.join(__dirname, 'copilotMetricsData.json');

Define the path to the local file where data will be stored

  let existingData = [];
  if (fs.existsSync(dataFilePath)) {
    const fileContent = fs.readFileSync(dataFilePath, 'utf8');
    existingData = JSON.parse(fileContent);
  }

Read existing data from the file, if it exists

  const newData = copilotUsage.filter(entry => !existingData.some(existingEntry => existingEntry.date === entry.date));

Filter out the new data that is not already in the existing data

  if (newData.length > 0) {
    existingData = existingData.concat(newData);

Append new data to the existing data

    fs.writeFileSync(dataFilePath, JSON.stringify(existingData, null, 2));
    console.log(`Saved ${newData.length} new entries.`);
  } else {
    console.log('No new data to save.');
  }
}

Save the updated data back to the file

orgMetrics();

Call the function

// Import Octokit
import { Octokit } from "octokit";

// **Import modules for working with files**
import path from 'path';
import fs from 'fs';
import { fileURLToPath } from 'url';
import { dirname } from 'path';

// **Declare variables for working with files**
const __filename = fileURLToPath(import.meta.url);
const __dirname = dirname(__filename);

// Set your token and organization
const octokit = new Octokit({
  auth: 'YOUR_TOKEN'
});

const org = 'YOUR_ORG';

// Call the API
async function orgMetrics() {
  const resp = await octokit.request(`GET /orgs/${org}/copilot/metrics`, {
    org: 'ORG',
    headers: {
      'X-GitHub-Api-Version': '2022-11-28'
    }
  });

  const copilotUsage = resp.data;

  // **Define the path to the local file where data will be stored**
  const dataFilePath = path.join(__dirname, 'copilotMetricsData.json');

  // **Read existing data from the file, if it exists**
  let existingData = [];
  if (fs.existsSync(dataFilePath)) {
    const fileContent = fs.readFileSync(dataFilePath, 'utf8');
    existingData = JSON.parse(fileContent);
  }

  // **Filter out the new data that is not already in the existing data**
  const newData = copilotUsage.filter(entry => !existingData.some(existingEntry => existingEntry.date === entry.date));

  // **Append new data to the existing data**
  if (newData.length > 0) {
    existingData = existingData.concat(newData);

    // **Save the updated data back to the file**
    fs.writeFileSync(dataFilePath, JSON.stringify(existingData, null, 2));
    console.log(`Saved ${newData.length} new entries.`);
  } else {
    console.log('No new data to save.');
  }
}

// Call the function
orgMetrics();

Run the script locally

After running the script with node copilot.mjs, you should have a new file in your directory called copilotMetricsData.json. The file should contain data from the API response.

If you run the script again tomorrow, it should only save data for one new day to the file.

You can work with the data from the API to identify trends over the last 28 days or, if you've stored data from previous API calls, over a longer period.

Example

In the following example, we update the orgMetrics function to extract the total and average number of active and engaged users per week. We could then use that data to track changes over time. This example uses the data returned directly from the API, and doesn't require stored data.

New steps are annotated in bold.

JavaScript
async function orgMetrics() {
  const resp = await octokit.request(`GET /orgs/${org}/copilot/metrics`, {
    org: 'ORG',
    headers: {
      'X-GitHub-Api-Version': '2022-11-28'
    }
  });
  const copilotUsage = resp.data;

Call the API

  let userTrends ={
    week1: {
      days:0,
      activeUsers:0,
      engagedUsers:0,
    },
    week2: {
      days:0,
      activeUsers:0,
      engagedUsers:0,
    },
    week3: {
      days:0,
      activeUsers:0,
      engagedUsers:0,
    },
    week4: {
      days:0,
      activeUsers:0,
      engagedUsers:0,
    },
  };

Create an object to store data for each week

 for (let i =0; i<copilotUsage.length; i++) {

Iterate over the data

    const week = Math.ceil((i+1)/7);

Determine the week number (1-4) based on the index

    userTrends[`week${week}`].days += 1;
    userTrends[`week${week}`].activeUsers += copilotUsage[i].total_active_users;
    userTrends[`week${week}`].engagedUsers += copilotUsage[i].total_engaged_users;
  }

Increment userTrends for the current week

 for (const week in userTrends) {
  userTrends[week].avgActiveUsers = (userTrends[week].activeUsers / userTrends[week].days).toFixed(2);
  userTrends[week].avgEngagedUsers = (userTrends[week].engagedUsers / userTrends[week].days).toFixed(2);
  }

Calculate the average number of active and engaged users per day for each week, rounded to two decimal places

  console.log(userTrends);
}

Output to the console

// Call the API
async function orgMetrics() {
  const resp = await octokit.request(`GET /orgs/${org}/copilot/metrics`, {
    org: 'ORG',
    headers: {
      'X-GitHub-Api-Version': '2022-11-28'
    }
  });

  const copilotUsage = resp.data;

  // **Create an object to store data for each week**
  let userTrends ={
    week1: {
      days:0,
      activeUsers:0,
      engagedUsers:0,
    },
    week2: {
      days:0,
      activeUsers:0,
      engagedUsers:0,
    },
    week3: {
      days:0,
      activeUsers:0,
      engagedUsers:0,
    },
    week4: {
      days:0,
      activeUsers:0,
      engagedUsers:0,
    },
  };

 // **Iterate over the data**
 for (let i =0; i<copilotUsage.length; i++) {
    // **Determine the week number (1-4) based on the index**
    const week = Math.ceil((i+1)/7);
    // **Increment userTrends for the current week**
    userTrends[`week${week}`].days += 1;
    userTrends[`week${week}`].activeUsers += copilotUsage[i].total_active_users;
    userTrends[`week${week}`].engagedUsers += copilotUsage[i].total_engaged_users;
  }

 // **Calculate the average number of active and engaged users per day for each week, rounded to two decimal places**
 for (const week in userTrends) {
  userTrends[week].avgActiveUsers = (userTrends[week].activeUsers / userTrends[week].days).toFixed(2);
  userTrends[week].avgEngagedUsers = (userTrends[week].engagedUsers / userTrends[week].days).toFixed(2);
  }

  // Output to the console
  console.log(userTrends);
}

Run the script locally

After running the script with node copilot.mjs, you should see output in your terminal like the following.

{
  week1: {
    days: 7,
    activeUsers: 174,
    engagedUsers: 174,
    avgActiveUsers: '24.86',
    avgEngagedUsers: '24.86'
  },
  week2: {
    days: 7,
    activeUsers: 160,
    engagedUsers: 151,
    avgActiveUsers: '22.86',
    avgEngagedUsers: '21.57'
  },
  week3: {
    days: 7,
    activeUsers: 134,
    engagedUsers: 123,
    avgActiveUsers: '19.14',
    avgEngagedUsers: '17.57'
  },
  week4: {
    days: 6,
    activeUsers: 143,
    engagedUsers: 132,
    avgActiveUsers: '23.83',
    avgEngagedUsers: '22.00'
  }
}