The first thing that went wrong at Veriff was a number. Its early product was aimed at Estonia’s e-residents: people who could use the country’s digital infrastructure without living there. The founders imagined a potential ten million of them, many needing bank accounts. In their October 2020 fifth-anniversary interview, the number actually enrolled was nearly 73,000. A market can be wonderfully modern and still be too small.
- The job: check that a remote customer is the person their document describes.
- The trade-off: stop fraud without losing legitimate customers in the queue.
- The lesson: keep the useful technology; reconsider the market you built it for.
Kaarel Kotkas also recalled that Estonian banks made life difficult for e-residents. The anticipated demand disappointed; the underlying problem survived. Strangers still needed to establish trust at a distance. Veriff widened its ambition from one country’s experiment to businesses dealing with customers anywhere. That change is the most instructive part of its origin: the founders could revise the customer without abandoning the capability.
01 / The photograph was too polite
Kotkas’s later account places another observation near the beginning. TransferWise, now Wise, had asked him to look at improving identity verification. Systems relying on static ID photographs were vulnerable to manipulation. A photograph offers evidence, but it does not object when someone edits it. The computer needs other ways to ask whether the document, the face and the situation belong together.
Founded in 2015 by Kotkas, with Janer Gorohhov as co-founder, Veriff launched its first product the following year. Its identity checks now combine document analysis, facial comparison and liveness detection. Depending on the product and plan, device, network and session signals add context; trained specialists handle cases through its hybrid service. The customer sees a short capture flow. The business receives a decision and supporting information.

The expertise is less glamorous than the word “AI” suggests. Documents differ across borders, generations and issuing authorities. Veriff’s published coverage includes more than 12,500 identity documents across 230-plus countries and territories. Knowing the legitimate variations matters: a system that mistakes unfamiliarity for fraud will punish exactly the international customers a growing business wants.
↓ Automated decision or specialist review
02 / A car key changes the question
Consider Bolt Drive. Its onboarding needs to establish that someone has an appropriate driving licence before giving them access to a car. In Veriff’s March 2025 customer interview, Bolt described preventing inexperienced drivers from using the service and blocking fraudulent users who tried to return with new accounts. Here, verification connects an online registration to a decidedly physical liability.
Veriff’s 2026 mobility report describes automated Category B licence-data extraction for Bolt Drive and a six-second average decision time. That is a customer-specific result, not a stopwatch promise for every applicant. Still, it explains the commercial appeal. A customer who came to rent a car should not need to reserve an afternoon for proving they can drive.
Trustpilot illustrates a different job. In June 2022, the companies announced optional verification of consumer reviewer profiles, with more than 50,000 consumers verified during the first months. A confirmed identity supplies a useful trust signal. It does not establish that the reviewer’s opinion is accurate, fair or worth reading. Businesses buying identity checks should be equally precise about what they expect the check to prove.
The customer list extends through financial services, marketplaces, gaming and online communities. LHV described fraudulent attempts arriving as soon as its onboarding went live. Argentine e-signature business Signatura reported fewer authentication-related support requests after integrating Veriff. The common purchase is a way to reduce uncertainty without turning each new customer into a clerical assignment.
03 / Trust has a minimum monthly bill
Veriff sells this capability as B2B software, through APIs, web and mobile SDKs, and hosted verification pages. Its wider catalogue includes returning-user biometric authentication, proof of address, database checks, age assurance and anti-money-laundering screening. The last of these checks sanctions, politically exposed persons and adverse media; ongoing monitoring can surface later changes.
Published self-serve prices put Essential at $0.80 per verification with a $49 monthly minimum, Plus at $1.39 with a $99 minimum, and Premium at $1.89 with a $209 minimum. Plus and Premium include verification specialists alongside automation. Add-ons cost extra; larger customers negotiate enterprise pricing. The business decision therefore includes review depth and volume, not simply the cheapest number on the card.
For a small team, the hosted page provides a starting point without building the capture interface. Developers can integrate the browser SDK or use the API to create sessions and receive results. A sensible pilot would include supported documents from actual customer markets, deliberate fraud attempts and ordinary users on ordinary phones. Record where people abandon the process. Then decide which cases need a specialist. The experiment addresses the problem Veriff’s customers describe: speed matters only when the resulting decision is useful.
Investors financed the expansion: $69 million in April 2021, followed by $100 million in January 2022. The latter round valued Veriff at $1.5 billion. That figure belongs to that financing date. The company said the new money would support global growth, research, product expansion and commercial teams. A valuation is a price paid for expectations; the customer’s test happens in the onboarding flow.
04 / The useful second look
Veriff competes with Persona, Sumsub, Jumio, Entrust’s Onfido and others. Its pitch rests on broad document coverage, session-level fraud signals and the choice of automated or hybrid review. In April 2026, it announced FIDO Document Authenticity certification for Full Auto IDV, describing independent testing across 19 document types from 12 countries. That bounded evidence is more useful than an unqualified promise of accuracy.
“You should never feel that you know everything, there’s always something to learn.”
Kaarel Kotkas / founders interview, 2020
The product continues to widen. February 2026 brought a marketplace-suite upgrade and a Data Zoo partnership adding authoritative data. Veriff’s published culture also stresses accountability and execution. Those are sensible ambitions for a company whose customers need decisions they can explain, rather than another impressive-looking dashboard.
A reader can copy one practical habit from this story: measure both sides of verification. Track detected fraud alongside completion, support requests and time to decision. BVG Jelbi’s customer account describes that ongoing balance. Test with the documents and people your service actually encounters. Poor capture conditions, unsuitable coverage or a badly chosen review policy can undermine the experience; a verified identity alone cannot guarantee good behaviour. Veriff’s first forecast was wrong. Its more durable insight was that the moment before saying yes deserves careful design.