Maven Course Scraper

Extract cohort-based courses from Maven.com — title, instructor, cohort dates, price, student enrollment, ratings, description, format, and topic tags.

How it works

Maven is a Next.js SSR application. Every listing and detail page embeds a __NEXT_DATA__ JSON blob that contains the full course data — no JavaScript rendering required.

Phase 1 — Listing: The actor fetches the provided start URLs (e.g. https://maven.com/courses?sort=trending). Each listing page returns up to 99 courses server-side. The actor expands a plain /courses URL into all three sort variants (trending, new, best-selling) to maximise coverage. Course slugs are deduplicated across all listing pages.

Phase 2 — Detail: Each unique course detail page (/school-slug/course-slug) is fetched concurrently. Detail pages contain richer data: full pricing, cohort schedules, ratings, and instructor social links.

Input

Accepted start URL formats:

• https://maven.com/courses — expands to all 3 sort variants
• https://maven.com/courses?sort=trending
• https://maven.com/courses?sort=new
• https://maven.com/courses?sort=best-selling
• https://maven.com/courses/ai — category leaderboard (top 9)
• https://maven.com/courses/product

Output schema (per course)

{
  "url": "https://maven.com/shreyas-doshi/product-sense",
  "course_id": 9991,
  "course_slug": "product-sense",
  "school_slug": "shreyas-doshi",
  "school_name": "Shreyas Doshi",
  "school_verified": true,
  "title": "World-class Product Sense in Practice",
  "description": "...",
  "format": "full_course",
  "price_usd": 3000.0,
  "currency": "USD",
  "num_ratings": 0,
  "avg_rating": null,
  "enrolled_last_week": "28",
  "social_image_url": "https://...",
  "tags": ["Product", "Strategy", "For Product Managers"],
  "instructors": [
    {
      "name": "Shreyas Doshi",
      "headline": "Former PM leader at Stripe, Twitter, Yahoo",
      "linkedin_url": null,
      "twitter_handle": "",
      "custom_social_link": null,
      "image_url": "https://...",
      "previous_employers": ["Stripe", "Twitter"]
    }
  ],
  "next_cohorts": [
    {
      "cohort_id": 24597,
      "name": "May 2026",
      "start_date": "2026-05-02T16:00:00Z",
      "end_date": "2026-05-09T23:00:00Z",
      "max_size": null
    }
  ],
  "next_cohort_start": "2026-05-02T16:00:00Z",
  "next_cohort_end": "2026-05-09T23:00:00Z"
}

Courses that fail to scrape are still pushed with an "error" field so the overall run always produces a complete dataset.

Notes

• Maven returns up to 99 courses per listing page server-side. The "Show more courses" button on the UI loads additional courses via JavaScript — the scraper does not execute JS, so it picks up the first page of each sort variant (~99 × 3 = ~297 unique courses before dedup).
• Category pages (e.g. /courses/ai) only return the top 9 leaderboard courses server-side.
• avg_rating is derived from sum_ratings / num_ratings / 10 to convert Maven's internal 100-point scale to a 10-point scale. Many courses have num_ratings = 0 if they use an external review platform (e.g. testimonial.to).
• No authentication is required. All data is publicly accessible.