Where this data comes from

Product catalog

Names, websites, descriptions, public pricing bands, and feature flags come from vendor sites, pricing pages, docs, and Afterstack editorial QA. Vendors can claim a profile to submit corrections and optional screenshot URLs. We do not invent list prices. Confirm current packaging on the vendor site.

Scores and quizzes

Attribute scores are Afterstack’s own 1–10 rubric, not imported star ratings. Live quizzes (CRM, Support, Sales engagement, Recruiting, Marketing, Knowledge) score fit from 0 to 100 against a perfect match, hard-filter, then move on those attributes. Affiliate payout, claimed profiles, migration-complexity labels, and product media are forbidden in the ranker. Category standing (#4 of N in CRM) uses that same fit model on a neutral profile, with every product in the category still in the list. The ↑ ↓ → mark compares today’s order to content/category-rank-snapshots.json. The first snapshot is a baseline, so the arrow stays off until a later catalog change moves a rank. Scoring methodology. Buyer-facing version: how Afterstack scoring works.

Company intelligence

HQ and founded year may come from the catalog briefing. Funding headlines, ownership, and M&A events are included only with a named source — company newsroom, investor relations, a company-issued wire, or a primary filing such as an SEC 10-Q. We do not scrape professional networks or invent valuations. If a newsroom does not publish a price, we omit it rather than fill it from secondary coverage.

The Airtable × Bending Spoons pages cite those companies’ August 2026 newsroom posts. The homepage and companies index “Recent company moves” module is a dated subset of the same curated records — editorial M&A and funding, not imported review text. Ticker-listing status (“public company on NYSE”) is a snapshot, not a news event, and is excluded from that rail.

Product media

Screenshots on product and company pages come only from Afterstack-hosted files under /product-media/ or vendor-submitted HTTPS URLs with a display license. Submissions from the claim page are queued unpublished in the same private Blob store reviews use when BLOB_READ_WRITE_TOKEN is set, and in content/product-media.json once an editor publishes them or commits the downloaded seed. They are labeled vendor-submitted and are not shown until that review. We do not scrape third-party screenshot galleries or invent product UI images.

Reviews

Written reviews are first-party Afterstack submissions via Leave a review. Each POST appends to a JSON store without the email, default published: false. On Vercel that store is private Blob when BLOB_READ_WRITE_TOKEN is set; otherwise it is an ephemeral file. The reviews hub, product pages, and /reviews/{slug} render only published rows (stars, first name, date, why-left / cutover / AI-trust excerpts). Until a review is approved, the product shows “Be the first to review” — never a fake user quote. Operators publish or reject from /admin/reviews using REVIEW_MODERATION_TOKEN or ADMIN_SECRET. Email is stripped from the JSON store and is never rendered. Product-page questions use the same log and webhook path and stay unpublished until editorial review. Published answers on product pages are Afterstack-authored FAQs in content/product-questions.json— first-party, not scraped. We do not import review corpora from other marketplaces. Vendor-submitted quotes must be labeled if they ever appear.

What we refuse

• Invented funding, deal value, or pricing.
• Paid placement as a rank or inclusion input.
• Putting migration specialists into software scores.
• Copying another directory’s category tree wholesale.
• Training or filling Afterstack from scraped third-party review text.
• Filling a company page from another software-review marketplace.
• Hotlinking vendor screenshots without a display license on file.
• Scraping third-party screenshot galleries or inventing product UI images.