Roblox's New Discovery Algorithm Reads 28 Days, Not One Week
The new window and the two entry-point signals explain how a low-pressure game like Brookhaven can sit in the top 10 for revenue while the algorithm stops rewarding day-one spikes.
Reporting from 1 source: 4Gamer.net.
At the Roblox Developers Conference 2026, Roblox staff and outside creators explained how the platform's recommendation system now works. The old algorithm judged games on a single 7-day window after launch. The replacement measures Day 1, Day 2-7 and Day 8-28 separately, and adds two entry-point signals: how often a Home recommendation is actually played, and how quickly first-time players leave. Brookhaven RP, which caps lifetime spending under 70 dollars, is in the top 10 for revenue across Roblox.
The discovery session had Behnam Rezaei, Roblox's vice president of engineering, and Sandeep Rajagopal, product manager for discovery features, walk through the change. Rajagopal said the 7-day goal of finding long-played games quickly was right, but the short window led to many misreadings.
The signals themselves did not change: play days, play time, co-play days, valid play sessions, monetization days and Robux spent. What changed is how long they are measured over. More than 20 signals feed the system, and Rajagopal said lining up that many is what lets different genres be compared on one footing.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.