Perplexity's Photon Paper Reveals Its Ranking Algorithm — a Playbook for Getting Cited
gaganghotra_ · x · 2026-10-20
Ryan Jones breaks down Perplexity engineering's Photon post (building a retrieval and ranking engine from scratch). Most coverage focused on speed — p99 retrieval-and-ranking latency dropped from 800ms to 65ms — but the bigger news for SEO/GEO practitioners is that Perplexity detailed exactly how it decides which pages to retrieve and rank before the model writes an answer.
Key points:
- Photon isn't one model "reading the web": each index shard retrieves candidates, runs an initial ranking, then a second-stage ranking, merging the best into the short list the answer model sees. If your page fails the first cut, you get zero visibility.
- The first pass looks far more like a classic search engine than "AI magic": it rewards the right words in the right fields, without stuffing, and orders candidates by a page-level usefulness signal.
- This is effectively an official ranking playbook — bigger than the Google leak, since it explains not just what's measured but how the metrics are used.
The article also maps how SERPrecon measures these same signals to help content get cited in Perplexity and other LLMs.
More from Apps
- Suno Highlights Five Niche Features Most Users Don't Know About — suno · 2026-10-23
- 3 devs spent 2 years building an AI-assisted roguelike horror card game, demo on Steam — AlvinhoGames_ · 2026-10-23
- 3 flight-booking rules my AI agent checks that I'd never do by hand — Efistoffeles · 2026-10-23
- OmarchyTV beta brings DHH's Omarchy-style launcher to Android TV — alexcovo_eth · 2026-10-23
- Perceptic launches PercepticOS, an AI workspace for 'scientific superintelligence' — nathanbenaich · 2026-10-23
- Capy 0.5.0 lets Claude Pro/Max subscribers reuse their plan quota, adds OIDC cloud access — garrytan · 2026-10-23