UntoldAi signals are built on five transparent pillars. No black boxes, no pay-to-rank. Here's exactly how we measure what's winning in the AI tools space.
Every upvote on UntoldAi is cast by a real, registered builder — not a bot, not a purchased click. When a builder upvotes a tool, that signal is logged with a tamper-proof timestamp, deduplication hash, and user fingerprint.
Raw averages are easy to game. A single 5-star review shouldn't beat a tool with hundreds of 4.8-star reviews. We use a Wilson score combined with a Bayesian prior mean to normalize ratings across tools with different review volumes.
Beyond votes, we track meaningful engagement: website visits, comments, follows, and reviews. Each action contributes points to a tool's signal score, capped daily to prevent engagement farming.
An AI tool that hasn't been updated in months should not perpetually outrank an actively developed one. The freshness signal applies a time-decay multiplier to each tool's score based on recency of product activity.
Trend signals measure the velocity of votes and engagement within a category over a sliding time window — not just total volume. A category with 50 new votes in 24 hours ranks higher than one with 200 total votes but no recent activity.
All public signals on this hub show aggregated data. No individual user's voting history, engagement patterns, or personal identifiers are ever published or exposed.
We collect engagement signals to power rankings, but we never sell individual data or expose per-user behavior publicly. Signal scores are computed server-side and published as aggregated category or tool-level metrics only.