Badges

Badges turn raw scores into meaningful labels, automatically classifying users as "Trusted", "Suspicious", "Bot", or whatever categories fit your business. They form the human-readable layer on top of numerical scores, and the foundation for automated decisions.

What Are Badges

A badge is an automatic label applied to an identity when its scores match conditions you define. Scores are continuous numbers that require interpretation; badges are discrete classifications that you can act on immediately.

Think of badges as the conclusions your team would draw from the scores, codified into rules that apply themselves. Instead of a team member seeing a Humanity score of 12 and thinking "that's a bot," the "Bot" badge makes that classification automatically, consistently, and instantly.

Badge Types

Every badge has a type that indicates whether it represents a positive, negative, or neutral classification:

• Good — positive classifications indicating trustworthy behavior. Examples: "Verified", "Trusted", "Power User". Good badges signal that an identity has passed your quality bar and may deserve elevated privileges or reduced friction.
• Bad — negative classifications indicating risk or abuse. Examples: "Suspicious", "Bot", "Freeloader", "Multi-Account". Bad badges flag identities that warrant restrictions, manual review, or automated action.
• Neutral — informational classifications that provide context without judgment. Examples: "New User", "High Volume", "International". Neutral badges help you segment your user base without implying risk.

The dashboard color-codes badge types so your team can spot patterns at a glance without reading every label.

Badge Rules

Badge rules define when a badge is assigned to or removed from an identity. Create rules in the dashboard under Settings. Each rule maps a set of conditions to a specific badge.

Score Conditions

The most common condition assigns a badge when one or more scores fall within a range.

• All four scores above 80 — assign "Trusted"
• Humanity below 25 — assign "Bot"
• Uniqueness between 20 and 50 — assign "Suspicious"
• Behavior below 30 AND identity older than 7 days — assign "Anomalous" (enough data to be confident)

Score ranges yield precise classifications. A Humanity score of 45 might not warrant a "Bot" badge, but it could trigger a "Possibly Automated" badge that puts the identity on a watchlist without drastic action.

Identity Age

Restrict badge rules by identity age. This supports time-bounded classifications like "New User" (first 48 hours), and ensures you only badge identities once enough data has accumulated for a confident assessment.

Combining Conditions

Rules can combine multiple conditions. All conditions must be true for the badge to apply. This yields highly specific classifications:

• Humanity above 90 AND Authenticity above 85 AND Uniqueness above 80 AND identity older than 7 days — "Verified" (high confidence across the board, with enough history to be sure)
• Uniqueness below 40 AND Authenticity below 50 — "Freeloader" (low uniqueness combined with disposable data strongly suggests trial abuse)

Badge Delays

Scores fluctuate. A user's Behavior score might dip to 45 for an hour, then recover to 72 as more events arrive. Without safeguards, badges would flicker on and off with every minor score change, creating noise and unreliable classifications.

Dregs prevents flicker with configurable delays on both applying and removing badges:

• Apply delay — how long the conditions must hold before the badge applies. A 1-hour apply delay means the scores must stay in the triggering range for a full hour before the badge appears.
• Remove delay — how long the conditions must remain unmet before the badge disappears. A 24-hour remove delay keeps the badge for at least a day even if scores briefly recover.

Hysteresis turns badges from volatile labels into stable, reliable classifications. An applied badge means the conditions were sustained, not momentary.

Badges and Alerts

Badges and alerts work together. Alert rules can use badge presence or absence as trigger conditions, creating a layered system:

1. Scores update continuously as events arrive
2. Badge rules evaluate those scores and apply stable labels
3. Alert rules trigger based on those badges and notify your team or systems

This chain — score triggers badge, badge triggers alert, alert sends notification — gives you the best of both worlds. Badges absorb the volatility of raw scores, and alerts act on settled classifications. The result: fewer false alarms and more actionable notifications.

For example, instead of alerting when Uniqueness drops below 40 (which might flicker), create a "Multi-Account" badge with a 2-hour apply delay, then alert when the badge appears. By the time the alert fires, the condition has been sustained.

Use Cases

Free Trial Abuse

Create a "Freeloader" badge that applies when Uniqueness is below 40. Set a remove delay of 7 days so the badge persists even if the user deletes their duplicate account. Pair with a Warning-level alert to notify your team for manual review, or a webhook alert to restrict the account automatically in your application.

VIP Users

Create a "Trusted" badge that applies when all four scores are above 80, with a 7-day apply delay. This identifies your most legitimate, long-term users. Use the badge to reduce friction — skip CAPTCHA, auto-approve actions, or offer premium features.

Suspicious Signups

Create a "Suspicious" badge for identities less than 48 hours old with Authenticity below 45. This catches disposable emails, fake names, and inconsistent signup data. Route these identities to a manual review queue before they can cause damage.

Bot Flagging

Create a "Bot" badge when Humanity is below 20, with a short apply delay and a long remove delay. Bots reveal themselves within minutes, but the label should stick for investigation. Pair with a Critical alert for immediate team notification.

Viewing Badges

Badges appear directly on identity cards in the dashboard, giving your team instant context when reviewing a user. Type a badge name in the search bar to see all identities carrying that classification.

Badges plus the search bar answer questions like "show me all users flagged as Suspicious this week" or "how many identities currently hold the Trusted badge."