Software Review Scraper — One Clean Schema (PeerSpot)

Honestly, I built this because I got tired of the same problem.

You want to know what real users say about Jira vs. Asana vs. ClickUp. So you scrape a few review sites, and now you have three different JSON shapes, dates in three formats, reviewer info buried in different places, and a pile of half-structured text you have to clean before it's useful to anything — a dashboard, a spreadsheet, or an LLM.

This actor skips that step. Every record comes out in one normalized format, no matter what it was scraped from:

• reviewer role and company, split into their own fields
• rating, title, and full review text
• the Q&A sections PeerSpot reviews are made of ("What is our primary use case?", "What is most valuable?", "What could be improved?") kept as structured question/answer pairs instead of mushed into one blob
• ISO dates (the real ones — I parse "14 May 2026", you don't have to)
• a stable review ID, so nightly runs dedupe for you
• a per-product summary record (aggregate rating, total review count) — diff that between scheduled runs and you get live rating-velocity tracking for free

What it costs

$2.00 per 1,000 results. A typical competitive-intel run (your product + 4 competitors, ~80 reviews each) lands around a dollar. Free-plan users can test it for pennies, which is how it should work.

Usage

{
  "products": ["Jira", "Asana", "https://www.peerspot.com/products/clickup-reviews"],
  "maxReviews": 100,
  "maxPagesPerProduct": 20,
  "includeQnASections": true
}

Plain product names are fine — they're resolved against PeerSpot's full catalog (23,000+ products). Slugs and full product URLs also work. If a product can't be resolved or a page fails, you get an honest product_error record telling you which one, not a silent gap.

Example output

{
  "record_type": "review",
  "review_id": "c9d10ee3f1293f98",
  "source_platform": "peerspot",
  "product": { "name": "Jira", "slug": "jira", "vendor": "Atlassian" },
  "reviewer": { "role": "SDE 2", "company": "Virtusa" },
  "rating": 4.5,
  "title": "Centralized sprint boards have transformed our planning...",
  "sections": [
    { "question": "What is our primary use case?", "answer": "My main use case for Jira involves..." },
    { "question": "What is most valuable?", "answer": "..." }
  ],
  "date": "2026-05-14",
  "source_url": "https://www.peerspot.com/products/jira-reviews?page=1",
  "scraped_at": "2026-10-24T21:36:16+00:00"
}

The full schema is published (schema.json) — validation errors on my side, not yours.

Who this tends to be for

• Competitive intel — one scheduled weekly run over your product + rivals; watch aggregate-rating deltas and mine new reviews for pain points people mention about your category.
• LLM / AI-agent data — the Q&A sections are already segmented prompt-shaped text; this is what they're good for.
• Sales — pull a competitor's 3-star-and-below reviews and you have an objection list that writes itself.

Straight talk about limitations

• v1 covers PeerSpot. G2, Capterra, and GetApp block datacenter traffic hard; they're on the roadmap behind this same output format, not promised this week.
• PeerSpot review volume is thinner than G2 for long-tail products — you'll get 5 reviews for some, 300 for the big names. product_error and review_count_total tell you what you're working with.
• Public review pages only, pacing built in, in line with the site's robots.txt. Nothing behind logins, nothing personal beyond what reviewers publish themselves.

Questions or a product that won't resolve? Open an issue or message me — I actually fix things fast on this one.