Last updated: September 2026
The best onboarding software for banks is the layer that runs a bank's identity checks rather than one of the checks. It orchestrates the KYC, KYB and identity verification providers the bank already uses, steps each applicant up to more checks only when their risk warrants it and down when it does not, and makes and explains every onboarding decision. It sends only the applications it cannot decide to a reviewer, with the context to decide them, and it leaves a record an examiner can follow. The reliable way to tell products apart on those points is to test them on your own past applications before signing.
This guide sets out eight criteria for choosing onboarding software, each with the questions to ask a provider and what a weak answer looks like, followed by a test to run on your own applications and a scorecard to reuse.
TL;DR
Choose the onboarding software that orchestrates your existing providers, steps applicants up or down by risk, decides and explains every application, and keeps the manual review queue to the cases a person needs to see.
Judge the software's approvals by what happened to the accounts afterwards, and not by pass rate alone. Approval rate is an appetite the bank sets.
The bank Customer Identification Program (CIP) rule, 31 CFR 1020.220, sets the floor the record has to evidence: name, date of birth, address and an identification number obtained before the account opens, with identity verified under risk-based procedures.
Oversight stays with the bank. The June 2023 interagency guidance on third-party relationships is in force and was proposed for replacement on 11 September 2026 (status as of 29 September 2026), and a bank's third-party programme covers the platform and every provider behind it.
Test each candidate by replaying your own past applications, and test a provider you have never used live on a slice of traffic.
Oscilar wrote this list and sells an onboarding and risk platform. The criteria name no providers.
The short answer: how a bank should choose onboarding software
A bank should choose the onboarding software that orchestrates its identity, KYC and KYB providers into one flow, steps each applicant up or down by risk, approves, declines or refers every application with a stated reason, keeps the manual review queue small, and leaves an examiner-ready record of each decision. The way to confirm those points is to replay a sample of your own past applications through each candidate before signing, because a demonstration shows what software can do in general and your applications show what it does for you.
Onboarding software matters because a KYC check answers a narrower question than the one a bank has to decide. A KYC check confirms that a claimed identity exists and matches its documents at a moment in time. It does not establish intent, and it does not establish that the identity was not assembled specifically to pass the check. Deciding what to do with an applicant who passes is the onboarding software's job.
Oscilar wrote this list and sells an onboarding and risk platform. Oscilar orchestrates third-party identity, KYC and KYB checks and is not a data or verification provider itself. The criteria below name no providers or products, and a later section applies the same criteria to Oscilar.
This guide covers choosing onboarding software as a whole. For the separate layers of fraud detection at account opening, and what each layer can and cannot see, read our guide to the account-opening fraud prevention stack.
What onboarding software does that a single verification provider does not
Onboarding software is the decision and orchestration layer of account opening. It decides which checks to run on each applicant, with which provider, in what order, and what to do with the results. A KYC, KYB or identity verification (IDV) provider answers one question about one applicant, such as whether a name, date of birth and address belong together or whether a document is genuine, and onboarding software combines those answers with the bank's own data and policy to reach a decision on the application.
An applicant who passes every check can still be a fraudster. A senior director of fraud risk management at a payments company described a past employer that left KYC and KYB results out of its models altogether: "Even in my past experience, we actually did not use any of the kyc or kyb vendors in our models because we just had way too many instances where fraudsters either just blatantly passed all the kyc kyb checks." The checks still matter, but the decision about an applicant who passes them has to come from the layer above.
In practice, onboarding software covers five jobs that no single verification provider covers on its own:
Orchestration: calling each provider in the order the bank sets, and handling a provider that fails or times out.
Risk-based flow: stepping each applicant up to more checks, or down to fewer, depending on risk.
Decision: approving, declining, stepping up or referring each application, with the reason attached.
Review: a queue and a case view for the applications a person has to decide.
Record: what was checked, by which provider, what was decided and why.
Some of the checks onboarding software runs are purchases in their own right. Choosing a KYB provider for the business checks themselves is a separate evaluation with its own criteria, and so is choosing an identity verification provider on its own. Where synthetic identities are the main risk, evaluating synthetic identity fraud detection software is its own exercise. The same is true of evaluating deepfake detection tools where manipulated selfies and video are the risk.
If you are still setting the wider approach, including the applicant's journey through account opening, start with our guide to digital-bank onboarding strategy. This guide assumes that approach is set and turns to the software that carries it out.
The eight criteria, and how to use them
The eight criteria follow an application through onboarding: orchestrating the checks, stepping the applicant up or down, making the decision, handling exceptions, working the review queue, changing policy, keeping the record, and fitting the bank's systems and oversight. Each criterion has the same three parts: what it is, what to ask a provider, and what a weak answer looks like.
Ask every provider the same questions in writing, and test each answer on your own past applications rather than on demonstration data. The choice is worth getting right the first time, because onboarding software sits across the whole account-opening flow and is costly to replace. A product team lead at a payments company said they wanted to be sure before choosing, "because it's a huge integration and it's a huge initiative."
1. It orchestrates your providers instead of replacing them
What it is: Orchestration is the software's ability to call the providers a bank already uses, in an order the bank sets, and to turn their answers into one flow. Banks already split identity checks across providers: one bank described using one provider to verify phone and email and a different one for name, date of birth and address. A technology leader at a community bank said verification tools used to be black boxes, and that what they look for now is a configurable flow into which they can map any data source they want to reach during the KYC process. An SVP of financial crimes strategy at a large regional bank said they had hoped for an orchestration layer where many verification vendors plug in and the rules can be changed quickly.
Orchestration includes cascading, where the software calls a second provider only when the first cannot verify the applicant. A senior director of risk decision analytics at a payments company was looking for a provider to sit last in a cascade of two or three, each called only when the one before it fails, to raise the share of applicants verified without a person. Building that by hand is slow and brittle. A compliance director at a small-business neobank described spending months trying to build a waterfall across verification partners, and a day when part of the onboarding flow broke, an engineer had to step in, and every application went to manual review.
Pass rate depends on the data a bank sends as much as on the provider it sends it to. The same senior director of risk decision analytics said a very low pass rate had once turned out to be caused by the poor quality of the data being sent, and not by the provider. Software that shows which fields went to which provider, and what came back, lets a bank tell those two causes apart.
The aim of orchestration is a better decision, whatever the number of providers. A risk leader at a regional bank said: "I don't like having too many vendors attached to a singular process." A VP of payment risk, compliance and operations at a payments company weighed the other side: "I don't want to limit the number of vendors just for the sake of, you know, the compliance concern and then compromising on the, you know, quality of our decisions, right?" Orchestration settles that tension by putting one decision over however many providers the bank needs.
What to ask a provider:
Build our current flow in your software, with our current providers in it. Which of them can you call today, and how is a new one added?
Can our team change the order of providers, or add a cascade step, without an engineering release?
When a provider times out or is down, what happens to an application in flight?
For each application, can we see which fields went to which provider and what each one returned?
What a weak answer looks like: A fixed flow built around the provider's own checks, where adding or reordering a provider is a professional services project. A pass rate quoted with no view of the data that produced it is another warning sign.
2. It steps applicants up by risk, in both directions
What it is: Risk-based step-up means the software adds checks when an applicant's risk warrants it and removes friction when it does not. The trigger is risk, which can appear after a check has passed as well as when one fails. The senior director of fraud risk management at a payments company quoted earlier said: "I would want to trigger a step up when I have fraud concerns for onboarding, beneficiary, owner verification, even if they pass the kyc."
The checks themselves should change with the applicant. A technical product manager leading a KYB redesign at a payments company described the requirement: "We need verification flow to be dynamic. For example, we want to create different verification templates depending on the industry, risk level, country or other parameters. Based on these parameters, we should be able to trigger a specific verification template with a specific set of fields or checks."
Step-up works in the other direction too. A senior director of fraud analytics at a consumer lender said: "I think if we could identify very low risk applications that we could reduce the friction on from authentication perspective, I think that would help us." Blanket step-up has a cost. At one lending marketplace, applicants who did not qualify for a fast path were sent through every stipulation page even when the data to decide was already on hand, and many of those applicants dropped off.
A step-up is a state in the flow as well as an extra check. The senior director of risk decision analytics quoted under criterion 1 said that once an applicant is stepped up to a document-and-selfie check, the rest of the flow should pause until the applicant completes it, because running the remaining checks is pointless if the applicant never clicks the link.
Some genuine customers will always be stepped up. A head of card fraud risk at a consumer lender acknowledged that "there will be chances that we inevitably will have to step up genuine customers", so the step-up has to be easy to get through. The senior director of fraud risk management at a payments company described the goal: "But we're looking to make this experience as seamless is the better word than frictionless because it is friction but seamless from the standpoint of like, how easy it is for the user to get through it."
Risk-based, layered controls are what the FFIEC's 2021 guidance, Authentication and Access to Financial Institution Services and Systems, supports. The guidance says: "Layered security incorporates multiple preventative, detective, and corrective controls, and is designed to compensate for potential weaknesses in any one control." It also expects a risk assessment to support decisions about authentication techniques.
What to ask a provider:
Show an applicant who passed KYC being stepped up on another risk signal, and a low-risk applicant skipping a step.
Where does the flow pause while an applicant completes a step-up, and what happens if the applicant never returns?
Can we set different verification paths by product, applicant type or risk level without an engineering release?
Run our past applications through your default policy. Which of them would it have stepped up, and why?
What a weak answer looks like: A step-up triggered only when a check fails, or the same extra checks for every applicant who misses a fast path. A step-up that restarts the application is another warning sign, because it means the flow has no notion of where the applicant is.
3. It makes the decision, and you judge it on the quality of approvals
What it is: The onboarding decision is the software's output for each application: approve, decline, step up or refer to a person, with the reason attached. Software that only passes provider results through leaves the bank's team to make each call by hand. The practical test of explainability is whether each decision carries reason codes and attributes that the analyst working the case, and later the record, can read.
Judge those decisions by the quality of the approvals and not by pass rate alone. The senior director of fraud risk management at a payments company said: "So we definitely want to optimize not necessarily just on the approval rates but also on the quality of those approvals." A high verification rate can come from a loose definition of verified, so ask how each provider in the flow defines a pass, and follow approved accounts to see what happened to them.
The approval rate itself is set by the bank's risk appetite. A VP of risk transformation at a merchant acquirer said: "That's really a policy thing, right? ... It's an appetite. We accept this and this; if the merchants deviate from that, your approval rate is your approval rate." A provider that promises a higher approval rate without asking about the bank's appetite is describing a looser policy.
What to ask a provider:
For a sample of our applications, show each decision and the reasons behind it, as a reviewer would see them.
How does each provider in the flow define a pass, and can we see the evidence behind each result?
Can we follow approved accounts after onboarding, by segment, to see which approvals later turned out to be fraud?
Which decisions does the software make, and which does it leave to our team?
What a weak answer looks like: A single score or a pass or fail flag with no reasons. An approval rate offered as a product feature is another warning sign, because the rate is a policy the bank sets.
4. It handles exceptions without sending everything to a reviewer
What it is: Exception handling is what the software does with an application that does not pass cleanly, and not every failure belongs in the review queue. A risk lead at a remittance company said: "For id V failures, don't bring it for manual reviews. Because even if you bring it for manual reviews, there's really nothing we're going to do about it. In such an instance, the customer should be prompted to retry."
Some exceptions look identical on the surface and need a path that tells them apart. The senior director of fraud risk management at a payments company said that when the credit bureau returns nothing on an applicant, the applicant may be a genuine thin-file customer, young or new to the country, or someone who entered a real-looking made-up name and date of birth. The no-hit result is the same for both, so the software should route it to further checks that can separate them, rather than approve or decline on the no-hit alone.
What to ask a provider:
Which failures send the applicant back to retry on the spot, and which go to a person? Can our team change that split?
Show the path for an applicant with no bureau record, once for a genuine thin-file applicant and once for a made-up identity.
When a provider returns an error rather than a result, what does the applicant see?
What a weak answer looks like: Every failure goes to manual review, or every no-hit is declined. The first moves the cost of each exception onto a reviewer, and the second moves it onto good applicants.
5. It leaves a review queue your team can staff
What it is: The review queue is the set of applications the software cannot decide and sends to a person. Its size, and the time each case takes, decide how many people onboarding needs and how long good applicants wait. A senior director of fraud analytics at a consumer lender described the cost: "So our manual review process, we pause applications in real time, right? And, this hurts our, this hurts our funding rate on all the applications, that we manually review because we don't work them on Sundays."
The reviewer should see everything needed to decide on one screen, with the reason the case was created. The same senior director said the lender had not automated more because its data sources are siloed and independent of one another, so a reviewer has to visit several separate systems to work one application. Software that brings each provider's result, the bank's own data and the reason for the referral into one case removes that step.
Consistent policy application is part of the case for letting software make the decisions it can. A VP of risk transformation at a merchant acquirer said: "The humans are inconsistent in the application of the policy, right? ... I'm tired on a Wednesday and I'm hungover on a Friday, right? And that's probably where we'll get the little bit of markup in that, just having that consistency."
What to ask a provider:
For our sample, how many applications would go to review, and for what reasons?
Show a review case. Does it show each provider's result, our own data and why the case was created, in one place?
How are cases prioritized, and what happens to paused applications outside working hours?
When a reviewer overrides the software, where is that recorded, and can our risk team use it to adjust the policy?
What a weak answer looks like: A queue that lists applications with a score and nothing else, leaving the reviewer to open each provider's portal. A review volume that nobody can estimate before go-live is another warning sign.
6. Your risk team can change policy, and test the change first
What it is: Policy control is who can change onboarding rules, how fast, and with what evidence that the change works. If every change needs an engineering ticket, the flow falls behind the fraud it is meant to stop, and risk teams stop proposing changes. The risk team should be able to change a rule, a threshold or a step-up path itself, within controls the bank sets.
Every change should be tested before it goes live. Backtesting replays historical applications through the proposed policy to show what it would have decided. An A/B test runs the new policy on a slice of live traffic beside the current one, and shows how it performs on applicants the historical data does not cover.
What to ask a provider:
Who on our team can change a rule, and what approval does the change need?
Show a policy change backtested on our historical applications, with the decisions that would have changed.
Can we run a change on part of our live traffic beside the current policy before rolling it out?
How is each change versioned, and can we see which policy version decided each application?
What a weak answer looks like: Policy changes submitted as requests to the provider, or tested only after they go live.
7. It leaves the record an examiner will ask for
What it is: The examiner record is the evidence, for each application, of what was checked, by which provider, what the software decided and why, and who overrode it. The floor it has to evidence is the bank Customer Identification Program rule, 31 CFR 1020.220, which requires a bank to obtain at a minimum, before opening an account, the customer's name, date of birth (for an individual), address and an identification number, and to verify identity under risk-based procedures. A head of risk and compliance at a mortgage lender described the need in governance terms: "there's the governance piece that's quite important as well in terms of how, you know, have oversight of decisioning and evidencing, and how you give those out, you know, deliver that evidence, shall we say to the regulator".
Where each element came from is now part of the record. Under two exemption orders, from the OCC, FDIC and NCUA on 27 June 2025 and from the Federal Reserve Board on 31 July 2025, each with FinCEN's concurrence, banks under those agencies may obtain a customer's TIN from a third-party source rather than from the customer. The relief is optional, and the bank must still obtain the TIN before opening the account under risk-based written procedures, so the software should record the source of each identifying element.
The record should be retrievable without asking anyone for it. A BSA compliance QA director at a sponsor bank, whose fintech partners collect data and documents at onboarding, asked: "But how do we feel comfortable with not only the data that they're kind of collecting, but also the documentation that they're collecting?" For a bank that onboards through partners, the software should hold that record where the bank can see it.
For business applicants, the record also covers beneficial ownership. Under FinCEN's customer due diligence (CDD) rule, 31 CFR 1010.230, a bank identifies each individual who owns 25 percent or more of a legal entity customer's equity interests, plus one individual with significant responsibility to control, manage or direct it, and verifies their identities under risk-based procedures. How a KYB provider resolves those owners belongs to choosing a KYB provider, which is a separate evaluation.
The record supports the bank's KYC programme without replacing it. For the programme itself, read our guide to building a KYC compliance programme.
What the record should show for every application:
The identifying information collected, and where each element came from, including any TIN obtained from a third-party source.
Every check run, with the provider, the fields sent and the result returned.
The decision, the policy version that made it and the reasons behind it.
Any step-up, and what the applicant completed.
Any manual review, who decided and why, and any override of the software's decision.
For business applicants, the beneficial owners identified and how each one was verified.
What to ask a provider:
Pull the full record for an application decided months ago, without opening a support ticket.
Can we export the record in a form an examiner can read without your software?
For applications onboarded through a partner, where do the partner's data and documents sit, and can we see them?
What a weak answer looks like: A pass or fail flag for each check, with the reasons held in the provider's logs, or a record the bank can only get by filing a request.
8. It fits your stack, your contracts and your oversight programme
What it is: Fit is whether the software works with the systems, provider contracts and oversight the bank already has. The software should connect to the bank's core systems, its own data and its existing providers, without asking the bank to re-platform. A product manager at a large bank described one provider's pitch: "Their pitch was basically, well, if you re orchestrate your entire backend and then dump everything into our platform, then it can do miracles. And I'm like that's cool. But we're never going to do that."
Swapping one provider should not mean rebuilding the flow. A technical product owner for fraud and identity platforms at a large regional bank said: "Some vendors' applications are very seamless. Some may not integrate well. There have been scenarios where we realized that something is not integrating well. We had to drop it and move to something else." Software that treats each provider as a replaceable step lets the bank drop one without touching the rest.
Provider contracts are part of fit. A partnerships manager at a business-banking fintech asked whether it could pull data through a platform while keeping its own commercial agreements with the data resellers, and a technical product manager leading a KYB redesign at a payments company asked: "Do we need separate contracts with individual registries? Or just one contract with you?" Either model can work, as long as the bank knows which one it is signing.
Oversight stays with the bank. A compliance lead at a payments company said that adding a platform between the institution and its providers does not move oversight, because the institution's vendor oversight programme still has to cover the platform and every third-party provider behind it. The June 2023 Interagency Guidance on Third-Party Relationships: Risk Management is in force and was proposed for replacement on 11 September 2026 by the OCC, Federal Reserve, FDIC and NCUA; as of 29 September 2026 the replacement is a proposal, with comments due 16 November 2026. The proposal emphasizes tailoring third-party risk management to each relationship's assessed risk and the institution's size and complexity.
What to ask a provider:
Which of our systems and providers do you connect to today, and what would we have to change to go live?
If we replace one provider, what else in the flow has to change?
Can we keep our own contracts with providers, buy through you, or mix the two?
What will you give our third-party risk team for due diligence on you and on each provider behind you?
What a weak answer looks like: A go-live plan that starts with moving your data into the provider's platform. A provider that treats the providers behind it as outside the bank's oversight is another warning sign.
How to test onboarding software on your own applications
To test onboarding software, replay a sample of your own past applications through each candidate and compare its decisions with what actually happened to those applicants. The senior director of fraud risk management at a payments company explained why a demonstration is not enough: "You know, it's always nice when you go to a vendor website and you see all these pretty things that they could do. And then you test them and you're like, well, what do they actually do?"
Pull a sample of past applications with known outcomes. Include frauds, good customers and applications that went to manual review, so the test covers the cases that decide the software's value.
Replay the sample through each candidate beside your current flow. An AML officer at an asset manager described doing this with earlier tools, in a "trial period test environment where we were able to pass through some of our old data and compare it to the pass rates that we had with older tools."
Run a provider you have never used live on a slice of traffic instead. A new provider has no history in your data to replay. The senior director of fraud risk management said: "Because obviously it can't be a back study. So maybe something more qualitative or something that we kind of have to plug in and just evaluate from, you know, like abc test perspective."
Compare decisions, step-up rate, review volume and the quality of approvals. Pass rate alone rewards a loose definition of verified, so follow the approved accounts in the sample to see which of them turned out to be fraud.
Pull the record for a handful of decisions. Check that an examiner could follow each one from the data collected, through each check, to the decision and any override.
Change one policy rule and time how long it takes to test and ship. The exercise shows who can make the change, whether it can be backtested first, and how long it waits on someone else.
Where Oscilar fits, judged on the same criteria
Oscilar wrote this guide, so this section applies the eight criteria to Oscilar using only what Oscilar's published pages state. Oscilar is not a data or verification provider. It orchestrates third-party identity, KYC and KYB checks together with the institution's own data.
On orchestration and step-up (criteria 1 and 2), the page on consumer onboarding on Oscilar describes orchestrating KYC checks and enhanced identity checks, connecting an institution's own databases, and connecting third-party data sources through a partner marketplace. Oscilar adds device intelligence, behavioral biometrics and data enrichments to those checks, runs onboarding workflows that adapt to each applicant's risk profile, and offers step-up verifications that trigger enhanced due diligence and additional verification steps when they are needed. Business onboarding works the same way, with KYB checks in place of KYC checks.
On the decision, exceptions and the review queue (criteria 3, 4 and 5), Oscilar combines rules, supervised machine learning and unsupervised anomaly detection. Its AI-driven case management prioritizes onboarding cases, gives a holistic view of each applicant and explains in natural language why each case was created.
On policy change (criterion 6), a risk team can build workflows in a no-code and low-code interface, backtest a new onboarding workflow on historical data before deploying it, run A/B tests, and monitor false positives, approval rates and its own KPIs afterwards. Oscilar also has auto model monitoring covering data drift, feature drift and concept drift, plus outcome KPIs measured by segment.
On the examiner record (criterion 7), Oscilar keeps a full audit trail per decision with stored reasoning, and human-in-the-loop by design. On fit (criterion 8), Oscilar is one platform across onboarding, credit, fraud and AML, and the platform is specified for decisions in under 100 milliseconds.
Oscilar does not verify documents or run liveness checks itself. Those checks come from the identity verification providers it orchestrates, so ask which provider a flow would call for each check, and run the test above on Oscilar with the same sample as every other candidate.
Oscilar was named to Chartis's 2026 FCC50, its ranking of financial crime and compliance technology vendors, with category wins for Low-Code/No-Code Customization and Agentic AI Innovation. Oscilar is also a Nacha Preferred Partner for Account Validation, Fraud Monitoring, and Risk and Fraud Prevention. Neither is a ranking of onboarding software, and neither replaces testing Oscilar on your own applications.
The eight criteria at a glance
The table pairs each criterion with the question that tests it, the answer that should give you pause, and the test to run on your own applications. Fill it in once for each provider on your shortlist.
Criterion | What to ask a provider | What a weak answer looks like | How to test it on your applications |
|---|---|---|---|
1. Orchestrates your providers | Build our flow with our current providers. How do we reorder one or add a cascade step? | A fixed flow where every change is a services project | Replay the sample with your providers in their current order, then with one reordered |
2. Steps applicants up by risk, both ways | Show a KYC pass stepped up on risk, and a low-risk applicant stepped down | Step-up only on a failed check, or the same extra checks for everyone | Check which applicants in the sample were stepped up, and why |
3. Makes and explains the decision | Show each decision with its reasons, as a reviewer sees it | A score or a pass or fail flag, or an approval rate promised as a feature | Follow the approved accounts in the sample to their outcomes |
4. Handles exceptions | Which failures retry on the spot, and what happens to a no-hit? | Every failure sent to review, or every no-hit declined | Trace the failed-check and no-hit cases in the sample |
5. Leaves a queue you can staff | How many of our applications go to review, and what does a case show? | A score and nothing else, with reviewers opening several systems | Count the review cases and have a reviewer work a few of them |
6. Your risk team changes policy | Who changes a rule, and how is the change tested first? | Changes by request to the provider, tested after go-live | Change one rule, backtest it and time the release |
7. Leaves the examiner record | Pull the full record for a past decision | Reasons held in the provider's logs, or records only by request | Pull records for a handful of decisions and follow each one through |
8. Fits stack, contracts and oversight | What changes to go live, and what do we get for due diligence? | Go-live starts with moving your data into the provider's platform | Swap one provider in the test flow and see what else changes |
Frequently asked questions
What does bank onboarding software do that a single KYC or identity verification provider does not?
Bank onboarding software decides what to do with an applicant, while a KYC or identity verification provider answers one question about that applicant. The software calls each provider in an order the bank sets, steps the applicant up or down by risk, and combines the results with the bank's own data and policy. It then approves, declines, steps up or refers the application with a reason, and keeps the record of each decision.
When should onboarding software step an applicant up to more checks?
Onboarding software should step an applicant up whenever the applicant's risk warrants it, including after a passed KYC check, and not only when a check fails. It should also step the lowest-risk applicants down to less friction. While an applicant completes a step-up, the rest of the flow should pause, and the step-up should be easy for a genuine customer to get through.
What should happen to an application the software cannot decide?
An application the software cannot decide should go to a reviewer with everything needed to decide it on one screen: each provider's result, the bank's own data and the reason the case was created. A failed selfie or document check is usually better handled by prompting the applicant to retry than by sending it to that queue. An applicant with no bureau record needs further checks that separate a genuine thin-file customer from a made-up identity.
What will an examiner ask to see from a bank's onboarding decisions?
An examiner reviewing onboarding will look for evidence that the bank met its Customer Identification Program under 31 CFR 1020.220: the name, date of birth, address and identification number obtained before each account opened, and how identity was verified under risk-based procedures. A useful record shows, for each application, what was checked, by which provider, what was decided and why, and who overrode the decision. For business customers, it also shows the beneficial owners identified and how each one was verified.
Can a bank get a customer's TIN from a data source instead of the customer?
Yes, if the bank is supervised by an agency that issued an exemption order. Orders from the OCC, FDIC and NCUA on 27 June 2025 and from the Federal Reserve Board on 31 July 2025, each with FinCEN's concurrence, let those banks obtain a customer's TIN from a third-party source rather than from the customer. The relief is optional, the bank must still obtain the TIN before opening the account under risk-based written procedures, and the text of 31 CFR 1020.220 was not amended.
How long should a bank test onboarding software on its own applications?
A bank should test onboarding software for as long as it takes to cover applications whose outcomes are known, including frauds, good customers and cases that went to review. A replay of historical applications can run as soon as the candidate is connected, because those outcomes are already in the bank's data. A live test of a new provider on a slice of traffic has to run until the accounts it approved have had time to show how they behave, which depends on the bank's products and volumes.
How does Oscilar measure up against these criteria?
Oscilar orchestrates third-party identity, KYC and KYB checks together with an institution's own data, and is not a data or verification provider, so it does not verify documents or run liveness checks itself. It offers onboarding workflows that adapt to each applicant's risk, step-up verifications, AI-driven case management that explains why each case was created, and backtests and A/B tests of policy changes. It keeps a full audit trail per decision with stored reasoning, and human-in-the-loop by design. Because Oscilar wrote this guide, test it on your own applications as you would any other candidate.
The best onboarding software for a bank is the one that performs on the bank's own applications. It orchestrates the providers the bank already uses, steps applicants up and down by risk, decides and explains each application, keeps the review queue small and leaves a record an examiner can follow. Oscilar's approach to consumer onboarding is one option to put through that test alongside any other candidate.

Oscilar Team
The Oscilar Team is comprised of experts from many domains of risk operations. These articles express viewpoints and knowledge from a variety of sources and contributors across the organization.
DISCLAIMER
The content on this website is provided for informational purposes only and does not constitute legal, tax, financial, investment, or other professional advice. Any views or opinions expressed by quoted individuals, contributors, or third parties are solely their own and do not necessarily reflect the views of our organization.
Nothing herein should be construed as an endorsement, recommendation, or approval of any particular strategy, product, service, or viewpoint. Readers should consult their own qualified advisors before making any financial or investment decisions.
Oscilar makes no representations or warranties as to the accuracy, completeness, or timeliness of the information provided and disclaims any liability for any loss or damage arising from reliance on this content. This website may contain links to third-party websites, which Oscilar does not control or endorse.


