Most identity checks announce themselves with the subtlety of airport security. Upload a license. Turn your head. Read a six-digit code. Answer a question about a street you lived on in 2007. Each step may catch a bad actor, but each also gives a perfectly good customer another chance to quit. Prove built its business inside that uncomfortable trade. The New York company sells identity verification and authentication to banks, fintechs, retailers, insurers, healthcare systems, crypto companies and marketplaces. Its pitch is not merely that it can recognize fraud. It is that trusted customers should feel less of the machinery.
The machinery begins with an object nearly everyone already carries: a phone. Prove looks beyond the number itself to signals about possession, ownership, device, SIM changes, reputation and behavior. It then links those signals to tokenized identity attributes. A business can use that confidence to pre-fill an application, approve a login, recover an account or step up a risky transaction. The phone is not the identity. It is the keyhole through which Prove inspects a much longer history.
The form was the first enemy
Prove began in 2008 as Payfone, when the mobile economy still sounded like a prediction. The original name made sense around mobile transactions. It made less sense as the company moved deeper into identity. A former marketing executive joked that it had become “an identity company with an identity crisis.” In 2020, after a $100 million investment and the acquisition of mobile-authentication assets from Early Warning Services, Payfone became Prove.
That rename matters because it captures what changed their mind. Payments were a use case. Identity was the infrastructure beneath many use cases. The company had already launched Trust Score and SIM Swap detection in 2015, then Pre-Fill in 2017. The latter turned a security signal into a growth feature: establish that the user controls a reputable phone, retrieve trusted identity information and populate the application. Instead of asking customers to prove themselves by typing more, the product tried to let verified data do the typing.
“We were looking to solve two primary challenges: reducing fraud and improving the customer experience by reducing friction.”Brandt Smallwood, Head of Card at Bilt Rewards
Bilt's public case study is the tidy version of the idea. The credit-card program said Prove helped validate phone ownership and possession for more than 90 percent of users, pre-filled half the form, reached a 90-plus-percent pass rate and took 48 hours to implement. A separate card-issuer case study reported 35 percent lower registration drop-off, onboarding that was 22 percent faster and an 87 percent instant-verification success rate. These are vendor-published customer results, not universal promises, but they show the scorecard Prove wants buyers to use: attack rate on one side, completed applications on the other.
What customers reported
Selected outcomes from separate Prove customer case studies. Different customers, flows and denominators; bars are not a head-to-head comparison.
What failed first
The more revealing story comes from Spark Wallet. Its chief executive said the company tried liveness checks, face ID and document scanning. Users complained. Abandonment stayed high. Worse, bots still got through. The team changed the order of operations: Prove Pre-Fill for onboarding, Trust Score for reputation and Mobile Auth for possession. Spark later reported eliminating bot attacks in the measured flow.
That result is not an argument against documents or biometrics. Prove itself partners with biometric provider authID and supports stepped-up verification. It is an argument against making every customer perform the most expensive ritual before the system has used quieter evidence. The important move is orchestration: passive signals first, added friction when risk earns it, and fallbacks when the silent route cannot reach a confident answer.
A product line built like a widening lens
Prove did not stop at faster forms. It bought UnifyID in 2021 to add passive behavioral biometrics and push authentication, then launched Prove Auth in 2022 for passwordless, cross-channel access. The Portabl acquisition in 2025 added reusable credentials, consent and interoperability. Unified Authentication pushed the pitch further away from the one-time code: recognize a trusted customer across sessions and channels, then respond when a phone number, device or carrier changes.
In October 2025, Prove introduced a Global Identity Graph that it says covers more than 90 percent of digital consumers across 227 countries and territories. The graph links Prove IDs to devices, national IDs, phone numbers, emails and authentication keys. In April 2026, the company gathered its products under the Prove Identity Platform: key management, Identity Manager and a global fraud policy. It now talks about verifying businesses and AI agents as well as people. ProveX adds real-time business verification through Baselayer, while the Agentic Suite aims to bind an agent's action to a verified human's consent.
This places Prove between familiar categories. Persona, Onfido, Jumio and Sumsub are associated with document and biometric proofing. Socure emphasizes identity and fraud intelligence. Alloy orchestrates decisions across data providers. Okta sits closer to workforce and customer access. A large bank may use several of them, plus its own risk engine. Prove's distinction is the phone-centric network and its attempt to preserve trust beyond signup. Its competition is not only another API. It is also the stack a company can assemble from bureaus, carrier checks, document scans, passkeys and SMS codes.
The business behind the invisible check
Prove sells enterprise software and API access through a sales-led model. There is no public menu price. That makes sense for a product whose cost depends on geography, volume, products and regulatory flow, though it also means a buyer must run a pilot to learn the economics. The company offers developer documentation, sandboxes, starter kits and public SDKs for TypeScript, Go, Java and .NET. Its own product page says a basic Pre-Fill setup can be deployed in minutes; a production rollout inside a bank is naturally a different creature.
Building this network was not cheap. Prove announced the $100 million investment in 2020 and another $40 million round in 2023. TechCrunch reported the latter valued the private company above $1 billion and that Prove was cash-flow positive at the time. The company says it now runs more than 30 billion verification events a year, holds more than 200 patents and serves thousands of businesses, including 19 of the top 20 U.S. banks. In 2026 it committed more than $5 million to expand its Ireland operation and add 50 roles in engineering, product, data science and research.
The customer list explains the model. Banks, card issuers, lenders and insurers lose money when fraud gets in, but also when good applicants give up. Healthcare systems need patient matching without turning account creation into paperwork. Retailers and gaming companies need to stop bot farms without interrogating every shopper. Prove sells into the budgetary seam between fraud loss, compliance cost, support cost and conversion. It works best when those numbers can be measured together.
The playbook worth stealing
Five moves to copy
- Find two teams with opposing KPIs. Prove gives fraud leaders stronger signals and growth leaders shorter forms.
- Use passive evidence before active challenges. Ask customers to work only when confidence is low.
- Return a visible benefit. Verification feels less extractive when it also fills the application.
- Measure a paired scorecard: fraud caught, good users passed, time saved and abandonment reduced.
- Expand along the same data advantage. Prove moved from onboarding to login, recovery, lifecycle monitoring and agents.
The copyable idea is not “use a phone number.” It is to make the security control pay rent in customer experience. Start with a narrow flow, establish possession before releasing personal data, compare pass rates and attack rates with a control group, and route uncertain cases to another method. If the pilot improves only security or only conversion, it has not resolved the conflict that made the product interesting.
Those conditions matter. A phone-first approach is less persuasive when the phone is shared, recently changed or disconnected from authoritative records. Silent checks can also confuse users if they fail without a legible recovery route. And persistent identity raises a governance question: the same continuity that makes fraud harder also makes consent, retention and correction more important. Prove describes its graph as permissioned and privacy-preserving; buyers still have to test that design against their own laws, customers and threat models.
“AI can fabricate a face or clone a voice, but it cannot replicate a decade of real digital behavior.”Rodger Desai, CEO and co-founder of Prove
That sentence is Prove's 2026 strategy in one line. Faces, voices and documents are becoming easier to synthesize. A long history of devices, keys and authenticated actions is harder to manufacture at once. Whether Prove becomes the trust layer for AI agents is still an open market question. Its more grounded accomplishment is already visible: it made a fraud-control company legible to people who care about completed forms.
The charming part is that the end state looks like almost nothing. A customer enters a phone number. The right fields appear. The risky session gets challenged. The familiar device passes. Security teams see a web of signals; the customer sees fewer chores. In enterprise software, disappearance can be a feature. Prove has spent 18 years learning how much business can fit inside it.