User quality score

Getting the score on webhook events

If you're using Neynar webhooks to get data on your backend, you might want to separate high quality data from low quality data. A simple way to do this is to look at the neynar_user_score inside each user object.

user: {
	fid: 263530,
	object: "user",
	pfp_url: "https://imagedelivery.net/BXluQx4ige9GuW0Ia56BHw/68c1cd39-bcd2-4f5e-e520-717cda264d00/original",
	profile: {
		bio: {
			text: "Web3 builder"
		}
	},
	username: "m00n620",
	power_badge: false,
	display_name: "tonywen.base.eth",
	experimental: {
		neynar_user_score: 0.9 // THIS IS THE SCORE
	},
	verifications: [
		"0xc34da1886584aa1807966c0e018709fafffed143"
	],
	follower_count: 29,
	custody_address: "0x22c1898bddb8e829a73ca6b53e2f61e7f02a6e6d",
	following_count: 101,
	verified_accounts: null,
	verified_addresses: {
		eth_addresses: [
			"0xc34da1886584aa1807966c0e018709fafffed143"
		],
		sol_addresses: []
	}
}

Fetching the score on API calls

If you're using APIs, you can get the same score as part of the user objects by passing in x-neynar-experimental boolean in the header. See screenshot below of User by FIDs for example.

Turning on this boolean flag will return the same score in the API or SDK response.

Interpreting the score

• A score above 0.95 has high probability of being a high quality user.
• Below 0.6 is quite likely to be lower quality.
• Between 0.6 and 0.95, it's ambiguous territory and depends on your application how you assess quality.

You can set the quality threshold on your product as per your preferences. Scores update few times a week so it's likely as new users become more active, their scores move higher.

Report errors

If you know a score is misrepresenting a user, that's helpful information we can use to label our data. Please send feedback to @rish on Warpcast DC or Telegram DM .