How Many Friends Does the Average Roblox User Have?

Why mean and median are wildly different

Roblox's friend-count distribution is heavily skewed. Most accounts have a small handful of friends — real-life classmates, siblings, the occasional player they met in-game. A small minority of accounts (creators, YouTubers, developers, the very-online) have hundreds. A tiny fraction have hit the 200 cap historically, and now the 1,000 cap (why the cap was raised).

When you compute the mean across all accounts, those high-end users drag the number up. When you compute the median, those high-end users count exactly the same as a kid with 3 friends. So:

• Median (typical user): ~10–20 friends
• Mean (sum-divided-by-count): ~40–60 friends
• Top 1%: 500+ friends
• Top 0.01%: at or near the 1,000 cap

The power-law distribution

This pattern — most users have few connections, a few have many — is called a power-law distribution and it shows up in almost every measured social network. Facebook, Twitter/X, LinkedIn, Instagram, GitHub, Discord — every one of them produces this same shape when you plot connection counts. It's so consistent that it's predicted by network-science models like the Barabási–Albert preferential-attachment model (essentially: people friend popular accounts in proportion to how many friends those accounts already have).

For Roblox specifically, the power law produces the hub accounts we wrote about in our hub-accounts post — the David Baszuckis, KreekCrafts, DenisDailys, Forstakens — whose disproportionate connection counts collapse the entire graph's average distance to ~4 hops.

How we know these numbers

FriendPath sees a sample of friend lists during searches. We aggregate friend counts (not who) into anonymized stats for these summaries. We don't store paths, and we don't connect any data to specific user IDs after the search completes. The shape we observe is consistent with everything publicly known about Roblox's userbase and matches the general power-law distribution found in other social-graph studies.

Implications for path searches

The median user having only 10–20 friends has real consequences for friend-path searches:

• Searches starting from a casual account often have a slower first BFS layer (50 first-hop friends gets you ~10 in practice)
• Hub accounts are the "highway system" of the graph; almost every successful long path routes through at least one
• Pairs of two casual users (10–20 friends each) often have longer paths (5–7 hops) than pairs that include a creator
• Users with 0 visible friends — banned accounts, UK-restricted accounts (why UK accounts hide friend lists in 2026), or accounts with private-friend-list settings — return "no path found" regardless of who the target is

The cap raise won't change the median much

The 200→1,000 friend cap raise affects mainly the top 1%. The median user has nowhere near 200 friends and won't have nowhere near 1,000 either — they're capped by their actual real-world social network. The cap raise lets hub accounts grow denser, which makes path lengths slightly shorter on average, but the typical-user experience is identical.