Change Tracking Methodology

Every day we save a copy of each tracked firm's pricing pages, rules pages, terms and every article in its public help center, and compare it with the last copy. When the text differs we record a change with the exact before and after.

• Help centers are listed through their sitemaps or public APIs (Intercom, Zendesk, Freshdesk, GitBook and others), so new and removed articles are caught too.
• We strip menus, footers, cookie banners and chat widgets, and ignore whitespace only differences.
• Our crawler identifies itself as PropBenchBot and follows each site's robots.txt. Pages behind a bot check are not crawled.
• Older history comes from the Internet Archive's Wayback Machine, and every change links to archived copies.

From October 25, 2026 every change comes from our own daily crawl. We saved our first copy of each page on October 24 and compare every new copy with the last one, so each change is dated to the day we saw it. Price and marketing pages must show the same new version on two crawls in a row before a change counts, which filters out live counters and rotating banners.

Before that, history is reconstructed from public copies saved by the Internet Archive's Wayback Machine. The archive only captures pages now and then, so we can say a page changed somewhere between two copies, not the exact day. These changes carry an Archive tag, show the date range, and link to both archived copies. Some firms are excluded from the archive and have no history before October 24.

Only importance 2 and 3 count toward a firm's nerf tally and the Rule Stability part of the score.