Fingerprint

The vendor says its free browser library identifies visitors at a materially lower rate than the paid server augmented service, so the accuracy gap is the model and the product. An identifier that survives cookie clearing, incognito browsing and virtual private networks for months cannot come from lookup alone.

The service returns an identifier and discrete signals for bots, virtual private networks and geolocation, and the customer's own code makes every decision. Nothing states the confidence attached to an identifier or how a borderline match should be treated.

Identification accuracy is stated at 99.5 percent with stability over months, a specific claim on the core capability. No methodology, population or measurement period supports the figure, though customers can validate identifiers against their own outcomes.

Named product managers at Uni Cards and Zebeede are quoted publicly, and a Canadian fintech reports roughly 70 percent of logins no longer needing a multi factor challenge. Marketplace reviews average 4.7 out of 5, but no request volume, customer count or fraud prevented figure is published.

Zero Trust mode resists requests impersonating another visitor, and a customer configurable filter limits which properties may call the service. Whether a device identified at one customer informs identification at another, the difference between a per customer tool and a global tracking graph, is never stated.

Separate data processing addenda for European and Californian law allocate the controller, processor and service provider roles explicitly, and a deletion process is documented. Published guidance separates legal bases by use case, telling customers that fraud prevention rests on legitimate interest while personalization likely needs consent.

A trust center publishes SOC 2 Type 2 and ISO 27001 with the Statement of Applicability, the scope document defining which controls the certificate covers. A penetration test report, nine governance policies and a four hour recovery time objective sit alongside it.

Privately held, the company claims no financial license, supervisory relationship or regulator counterparty. A software directory lists it as supporting payment card and second Payment Services Directive compliance, but its own material makes neither claim.

Shared, refurbished and older devices are harder to tell apart and skew toward lower income users, and a device wrongly matched inherits another visitor's fraud history. No false positive rate, coverage breakdown by device or fairness testing is published, nor any account of how collisions are detected or corrected.

The company will not bill requests generated during a denial of service attack and will waive charges if a malicious surge gets through, though identification quality carries no guarantee. A person whose device is matched to another's history has no appeal at the vendor, only the deletion route through the customer as controller.

The client side collection library is published as source with more than 28,000 repository stars, so an engineer can read exactly what is gathered from a browser. Server augmented matching, where the accuracy advantage comes from, discloses no data, models or external sources.

Client kits ship for web, Android and iOS among others, and edge integrations at Cloudflare, Akamai and Fastly let identification run through the customer's own first party origin. Segment and Google Tag Manager connectors and public developer documentation complete a catalog aimed at engineers rather than fraud analysts.

Data sits in Amazon Web Services data centers, named openly, but no region, country of processing or transfer mechanism is stated. No private or single tenant deployment of the commercial service appears, despite customers on four continents.

A free plan, Pro Plus at $99 a month for 20,000 web and iOS requests, overage at $4 per thousand and a 14 day unlimited trial make up a full rate card. A request bills only when it identifies a device and returns an event identifier, and cached responses never bill, which defines the unit precisely.

Dedicated material covers fintech, banking and buy now pay later fraud, and pricing spans a single developer on the free tier to contracts above nine million requests a year. The buyer is a developer adopting per use case rather than a fraud function, with no sales into banking operations or insurance evident.