UserGate
Early access · No credit card

Find the leak in your free tier.

One person, a dozen free accounts, and your compute bill pays for all of them. UserGate pinpoints the duplicate signups, trial farming, and shared logins draining your margins, and gives you one click to stop it, without banning real customers.

  • No credit card to start
  • Live in 15 minutes
  • Never bans real users
cluster_4827
multiple_accounts 0.94 · HIGH
mainacct_1acct_21.00acct_30.92acct_40.38
Free compute burned this month$3,100

12 free trials claimed · 4 linked accounts · 1 device

Recommended: pool into one free quota

Built for the stack you already ship on

StripeClerkSupabaseVercelNext.jsWorkOSAuth0PostgreSQLStripeClerkSupabaseVercelNext.jsWorkOSAuth0PostgreSQL

First-class drop-ins for your auth and billing layer: adoption is one toggle inside a tool you already trust.

The leak

Your free tier costs more than it should

It never shows up as a refund or a chargeback. It shows up as a slightly worse gross margin, an inflated signup count, and a free tier that quietly bleeds money.

Account duplication

One person spins up many accounts to keep claiming free credits, trials, and new-user perks.

Burns compute on every use

Account sharing

One paid login passed around a team or a group who should each be paying for their own seat.

Silent seat revenue, gone

Fake, bots & temp emails

Throwaway signups from disposable domains that pollute your data and drain free resources.

Inflated signups, worse margin

Why now

AI compute changed the math.

On a normal SaaS product, a free abuser costs close to nothing. On an AI product, every free generation burns GPU and API spend. Free-tier abuse went from an annoyance to a direct, visible hit on gross margin, felt immediately, in dollars.

What a founder actually sees

$3,100

of compute burned this month by 40 linked accounts.

A number you act on today, not a risk score you have to interpret.

How it works

Four layers. One outcome: revenue you stop losing.

The difficulty is lopsided, and the value is not where the difficulty is. We do the hard linkage work so you can act in one line of code.

01

Signal collection

drop-in snippet

A few lines of code, no impact on UX. Signals are gathered at signup and handed straight to the engine, never a black box in your login flow.

02

Entity resolution

the real core

Record linkage, not an LLM. Deterministic hard-links first, then probabilistic scoring, then graph clustering into one entity.

03

Usage anomaly

metered-specific

Unsupervised clustering over usage patterns catches the signature of credit farming even when the hard signals are thin.

04

Decision & action

highest value

A confirmed cluster plus a recommended action, surfaced as a dashboard, a webhook, and drop-in components you wire in.

Detection lives server-side. You make one call.

The matching and scoring brain runs on our servers. You send the signup's signals and get back a resolved cluster and a recommended action, so abusers never get the evasion playbook, and you never trust client-supplied data.

app/api/signup/route.ts
// 1. send the new signup's signals
const res = await fetch("https://api.usergate.dev/check", { userId })

// 2. get back the cluster + an action
const { cluster, action } = await res.json()

if (action === "pool_quota")
  grantFreeTier(cluster.entityId) // one quota, shared
The signals

We link accounts with the signals that actually survive

Ordered by value-per-cost. Deterministic hard-links first, probabilistic signals layered on top, and the payment fingerprint where the browser fingerprint runs thin.

Strongest · zero-cost

Stripe card fingerprint

Stripe exposes a stable fingerprint for the same card across accounts. Beating it requires a brand-new real card per account, which destroys the abuser's economics.

evidence

acct_1 card_fp: pm_1Nabc…f9

acct_2 card_fp: pm_1Nabc…f9 ← match

acct_3 card_fp: pm_1Nabc…f9 ← match

The product

Detection is table stakes. The action layer is the product.

Every other tool hands you a risk score and leaves you to wire up the consequences. We ship the playbook: a menu of soft, reversible, mostly user-driven remediations, ordered soft to hard.

  • Pool into one free quotaheadline
    use case: Usage
    Very low

    All linked accounts share one free allowance. No bans, ever. The headline action.

  • Step-up: require phone or card
    use case: Both
    Low

    Kills the economics of farming without blocking a single real user.

  • Consolidate email
    use case: Content dedup
    Low

    A friendly “merge into your main account” nudge. User-driven.

  • Upsell prompt
    use case: Sharing
    Low

    “Add a seat for $X.” Turns abuse straight into revenue.

  • Device-management flow
    use case: Sharing
    Low

    “Keep this device, remove that one.” The user chooses. Anchored on device, never IP.

  • Throttle
    use case: Usage
    Medium

    Quietly slows the abuser down instead of confronting them.

  • Hard block
    use case: Both
    High

    Reserved for repeat, blatant offenders only. The last resort.

Every action defaults to soft, reversible, and user-driven, so you are never accusing a paying customer of cheating. Inform first through dashboard, webhooks, and components; automate later, only when you want to.

Drop-in components

Pixel-perfect remediation, embedded in minutes

The action layer ships as drop-in React components that match your brand. Detection happens server-side; you render the fix. Pick one to preview it live.

app.yoursite.com
Shared free tier

You and 2 linked accounts share one free allowance.

credits used38 / 50
// <QuotaPool entity={cluster.entityId} />Explore all components
Quota pooling

One real person, one free tier.

The headline remediation. All linked accounts share a single free allowance: three accounts that link together get one free tier between them, not three.

Without poolingabuse
acct_150/50
acct_250/50
acct_350/50

150 free credits handed to one person.

With UserGateneutralized

entity_8842 · 1 shared quota

acct_150/50
acct_250/50
acct_350/50

50 free credits, as intended.

Tolerant of false positives by design. If we wrongly link two real users, the worst case is they share a free quota, and nobody gets locked out. That is the single most important property for shipping an imperfect engine you can actually trust.

AI-native, done right

An analyst that writes the case for you

The tedious human job in fraud is the investigation, not the math. So we point the AI where the labor is: classical ML scores the engine, and an LLM reads a flagged cluster, explains it in plain English, and drafts the action.

usergate · analyst
Why was cluster_4827 flagged?
3 accounts, shared device fingerprint, created within 6 minutes, each used exactly 50 free credits then went idle. 92% likely one person.
confidence0.92
draft · consolidation email

to: john@gmail.com

subject: Merge your UserGate accounts into one?

Hi, looks like you’re running a few accounts with us. Want us to merge them into your main one so your credits and history live in a single place? One click and you’re set.

Classical ML in the engineLLM in the analyst & the UXDrafts the email, answers “why flagged”
Everything in the box

Everything you need to plug the leak

Detection, the remediation playbook, and the components to ship it: one integration, no risk score left for you to interpret.

Quota pooling: the headline remedy

Linked accounts share a single free allowance. Three accounts that link together get one free tier between them, not three. No bans, ever. Even a false positive just shares a quota.

without pooling

150

free credits to one person

with usergate

50

free credits, as intended

Server-side by design

The matching brain runs on our servers, so abusers never get the evasion playbook and you never trust client data.

Stripe card fingerprint

The strongest, zero-cost signal. Beating it costs a brand-new real card per account, and that breaks the economics.

An AI analyst on the case

Classical ML scores the engine; an LLM reads the cluster, explains why it flagged, and drafts the action for you.

Drop-in integration

Add it in an afternoon: no black box in your login flow, no sales call, no procurement.

Dashboard, webhooks & components

Surface a confirmed cluster three ways: a dashboard to watch, webhooks to automate, and drop-in components to act. Inform first; automate only when you want to.

False-positive tolerant

Because the worst case is a shared free quota, you can ship an imperfect engine and still trust it: a bar a solo builder clears and a hard-block product cannot.

Who it's for

Built for the case where abuse is real money

We start where the ROI is instant and measurable, then expand into the platforms where the damage is a sale you never made.

start here

Usage & metered builders

Anything that gives away free credits that cost real money to serve. Every duplicate account burns GPU and API spend on every use, so the ROI of stopping it is immediate.

AI image & videoAI writing & chatAI agentsTranscriptionCode toolsMetered APIs
  • Detection at the moment credits are granted or consumed
  • Quota pooling as the native, no-ban remedy
  • “These 40 linked accounts burned $3,100” is a number founders act on
next

Content & access platforms

Courses, gated libraries, and paid communities. The damage is the sale you didn’t make: duplicate/trial-replay accounts, and one paid login shared across many people.

Course platformsGated librariesPaid communitiesMembershipsLicenses
  • Same dedup engine, with a soft-to-hard remediation ladder
  • Account sharing caught by device identity + concurrent sessions
  • Device-management flow: “keep this device, remove that one”

Never cap by IP. A household shares one IP (hiding real sharing); one real user touches many in a day (flagging your customers). We anchor on device identity and concurrent sessions, the most common mistake in the category, avoided.

Why UserGate

We don't win on detection depth. We win on what happens next.

Detection is a commodity: off-the-shelf fingerprinting is free. The value is the remediation, the integrations, and the niche we're built for.

The action layer

We ship the remediation playbook, not a risk score. Turnkey where the rest of the category is a toolkit.

Built for usage & AI abuse

The tool for when abuse is real compute cost, with quota pooling as the native remedy. Nobody else owns that framing.

Developer-first

Drop it in during an afternoon, no black box in your login flow. Self-serve kills the sales cycle.

False-positive tolerant

Quota pooling delivers value without perfect precision: a bar a solo builder can clear and a hard-block product cannot.

What you get
others: A visitor ID and a risk score
usergate: The duplicate cluster + a recommended action
What you do next
others: Build all the enforcement yourself
usergate: Ship the playbook in one line of code
Built for
others: Generic fraud, mid-market & enterprise
usergate: Usage-based & AI free-tier abuse
A false positive
others: Locks out a paying customer
usergate: Worst case: they share a free quota
Distribution
others: Sales-led demo and a contract
usergate: Self-serve, install in an afternoon

We catch the casual majority, not the professional ring. Device farms, emulators, and residential-proxy operations are a different fight, and we’ll tell you that to your face before you ever pay us.

Pricing

Start free. Pay when it pays for itself.

Product-led and honest. Fraud is continuous and adversarial, so this is a subscription that keeps working, not a one-time snapshot.

Free

$0/ month

See the leak. Built entirely from zero-cost signals.

Run the free audit
  • Hosted detection engine
  • Device + email + Stripe card fingerprint
  • The free audit & dashboard
  • Up to 2,000 tracked accounts / mo
  • Webhooks
Most popular

Pro

$49/ month

Stop the leak at scale, with the full action layer.

Start free trial
  • Everything in Free
  • Paid signals: proxy / VPN + phone OTP
  • Full action layer: pooling, step-up, device-mgmt
  • AI analyst: cases & drafted emails
  • Drop-in components + Stripe / Clerk / Supabase
  • Priority support

Scale

Custom

For high-volume platforms with their own rules.

Talk to us
  • Everything in Pro
  • Higher tracked-account limits
  • SSO, audit logs & custom retention
  • Custom signals & integrations
  • Dedicated support & SLA

Usage-aligned pricing on monthly tracked accounts · Team & growth tiers above · Paid-API signals gated so the free tier stays free to run

Questions

The objections, answered straight

No. Every action defaults to soft and reversible: quota pooling and step-up over hard blocks. A false positive should annoy a real user for a moment, not lock them out. Hard blocks are reserved for repeat, blatant offenders only.

The 15-minute audit

Find the revenue leaking out of your free tier.

Connect in 15 minutes. See what percent of your signups are linked duplicates, and exactly what they cost you this month. The reveal sells itself.

No sales call · No credit card · One real person, one free tier