The Uniqueness Score
Uniqueness score / noun / — a measure of whether an account is the only one its owner operates. A uniqueness score aggregates cross-account evidence like shared devices, shared IPs and sessions, suspiciously similar identity data, and behavioral overlap to answer the question every referral program, free trial, and voting feature depends on: is this one person?
Dregs computes a Uniqueness score from 0 to 100 for every user, continuously, as one of its four identity scores. It's the dimension built to handle multi-accounting, the abuse pattern where every individual account looks fine and the fraud is only visible across many accounts.
Device Fingerprinting and Other Cross-Account Signals
Uniqueness is the one score that can't be computed from an account in isolation. Every signal is relational, comparing the account against every other identity you track, starting with cookieless device fingerprinting:
How Dregs Computes the Uniqueness Score
A pipeline of analyzers runs the cross-identity queries (which accounts used this device, this IP, this session, this naming pattern) and produces observations with a value from 0.0 (suspicious) to 1.0 (legitimate), a confidence, and an explanation. The Uniqueness score is the weighted aggregate of the Uniqueness-category observations, scaled to 0–100. The same evidence also builds the identity graph, so every low score comes with a map of exactly which accounts it links to and why.
{
"category": "UNIQUENESS",
"id": "uniqueness.shared-device",
"label": "Shared Device",
"value": 0.2,
"confidence": 0.85,
"explanation": "Device fingerprint shared with 3 other identities",
"metadata": { ... }
}Why Multi-Accounting and Duplicate Accounts Slip Through
Signup-time defenses check accounts one at a time, and one at a time is exactly how multi-accounters win: unlimited email aliases, VoIP phone numbers for pennies, a VPN for every IP check. Each new account clears the bar because the bar never looks sideways at the accounts that came before.
The Uniqueness score looks sideways by construction. It's recomputed as evidence accumulates, so the third account from a device drags down all three, and the referral chain that traces back to one browser unwinds in the graph. See it applied in the duplicate accounts, referral fraud, and free trial abuse use cases.
The Other Three Scores
Uniqueness answers "how many accounts is this person running?" The other three dimensions profile what each of those accounts is.
Humanity
Is a real human controlling the browser, or automation?
Authenticity
Does their claimed identity look real, or fabricated?
Behavior
Does their usage look like a legitimate customer journey?
All four come from the same identity scoring pipeline, on every plan, from the events you're already sending.
Frequently Asked Questions
Q: What is a uniqueness score?
A: A uniqueness score is a continuous measure of whether an account is the only one its owner operates. In Dregs it's a 0–100 value computed from cross-account evidence: devices shared between identities, shared IPs and sessions, suspiciously similar names and emails, and behavioral overlap. A score near 100 means the account stands alone; a score near 0 means it almost certainly belongs to someone who also operates other accounts.
Q: What is multi-accounting?
A: Multi-accounting is one person operating multiple accounts on a platform that expects one account per person, whether to farm referral bonuses, stack free trials or resource limits, manipulate votes and reviews, or evade a ban. Each account looks normal in isolation, which is what makes multi-accounting hard to catch with per-account checks. Detection has to happen across accounts, which is exactly where the Uniqueness score operates.
Q: How does Dregs detect duplicate accounts without government ID or phone verification?
A: By evidence instead of credentials. Cookieless device fingerprinting survives incognito mode and cleared cookies, so the same physical browser shows up across accounts. Shared IPs and sessions, names and emails that vary by a suffix, and near-identical usage patterns each add their own signal. No ID checks, no phone numbers, no signup friction: the tells accumulate from ordinary usage.
Q: Won't shared devices cause false positives, like families, offices, and libraries?
A: Legitimately shared devices are why Uniqueness is a score, not a verdict. A household device shared by two real customers lowers Uniqueness somewhat, but both accounts keep strong Authenticity, Humanity, and Behavior scores, a very different profile from ten thin accounts cycling one device. You can also mark known-shared devices or trusted identities as disregarded, which removes them from cross-account analysis entirely and automatically re-scores anyone affected.
Q: What's the difference between the Uniqueness score and the identity graph?
A: The Uniqueness score is built in part from the identity graph. The identity graph is the map: which accounts link to which, through what (shared device, shared IP, similar email, similar behavior). The Uniqueness score is the per-account summary of that map, distilled to a 0–100 number you can sort, badge, and alert on. Investigate rings in the graph; automate policy on the score.
Q: Can I allow some users to have multiple accounts?
A: Sure. Plenty of products have legitimate multi-account users: agencies, resellers, or even your own staff. Mark those identities as disregarded and their devices and artifacts stop counting against anyone's Uniqueness, including their own. Policy stays yours: the score tells you who's operating multiple accounts; your badge rules, alert rules, and webhooks decide what happens about it.
One person, one account, enforced by evidence.
Drop the Dregs tracking script into your application and every identity gets a continuous Uniqueness score (plus Humanity, Authenticity, and Behavior) with the identity graph to back it up.
Schedule a Demo