Ongoing CDD software RFP: requirements and recommended vendors

On this page
Ongoing customer due diligence (perpetual KYC and KYB) is risk-based monitoring across the life of a business relationship. Onboarding is the first review, not the last. An RFP for this software has to test whether a vendor can keep customer and beneficial owner information current, detect trigger events, route alerts to reviewers, and produce records that satisfy a competent authority.
This document gives procurement and compliance teams a copyable RFP structure: must-have requirements, a question set by capability, a weighted scoring matrix, an evidence checklist, and contract clause prompts. It also answers the question of which vendors to shortlist.
What regulators expect from ongoing monitoring software
The European Banking Authority's Guidelines on ML/TF risk factors (EBA GL 2021/02, consolidated version) set out the obligations your RFP should map to. Five provisions matter most for software selection:
Guideline 1.28: firms "should use information obtained during the course of the business relationship for individual risk assessment purposes."
Guideline 4.70: monitoring includes transaction monitoring and "keeping the documents, data or information they hold up to date, with a view to understanding whether the risk associated with the business relationship has changed."
Guideline 4.71: firms "should determine the frequency and intensity of monitoring on a risk-sensitive basis."
Guideline 4.73: firms "should ensure that they have processes in place to review flagged transactions without undue delay."
Guidelines 5.1 and 5.2: firms must keep records of CDD information, risk assessments and transactions, and those records should be "sufficient to demonstrate to their competent authority that the measures taken are adequate in view of the ML/TF risk."
Guideline 4.76 is the shortest and the most operational: "Firms must keep CDD information up to date."
The FCA Handbook (FCTR 5.3, examples of good and poor practice) shows what risk-tiered review frequency can look like in practice. It describes a large firm that reviews clients rated A and B every three years, Cs every two years, and Ds and Es annually. Treat that as an illustration of the pattern, not a required schedule. Your software must let you configure whatever schedule your own risk assessment supports, and then prove it was followed.
Define the RFP objective before drafting questions
State one objective in the RFP preamble: identify vendors that can run continuous or trigger-based monitoring with defensible decisioning and auditable evidence. Every question below should trace back to it.
Three scoping decisions shape the rest of the document:
Entity scope. Decide whether the software covers business customers only (KYB), individuals only (KYC), or both, including beneficial owners and legal representatives.
Monitoring model. Decide which mix of real-time, trigger-based and scheduled periodic review you need. Most programs need all three.
Build boundary. Decide whether the vendor supplies data, workflow, or both. Data-only providers require you to build triage, case management and audit logging yourself.
For deeper clause language on enterprise and complex-structure scenarios, see the enterprise KYB vendor RFP clauses and rubric.
Must-have requirements for ongoing customer due diligence software
Use these as the mandatory requirements section of the RFP. Vendors should respond to each with product behavior, a measurable metric where one applies, and an evidence artifact.
Requirement | What the vendor must support | Regulatory anchor |
|---|---|---|
Risk-based monitoring model | Monitoring frequency and intensity configurable by customer risk tier and by trigger (ownership change, sanctions or adverse media hit, document expiry, behavior change) | EBA 4.71 |
CDD data coverage | Customer and beneficial owner data, registry connections, UBO register access, evidence standards per jurisdiction | EBA 4.70(b), 4.76 |
Continuous screening | Configurable sanctions, PEP and adverse media screening with defined frequency and re-screening on data change | EBA 1.28 |
Transaction monitoring integration | Ability to ingest or receive monitoring outcomes and feed them into the customer risk profile | EBA 4.70(a), 4.72 |
Alert triage and evidence collection | Reviewer queues, rationale capture, evidence attachment, escalation and four-eye review | EBA 4.73 |
Policy-driven decisioning | Rules and risk scoring defined as versioned policy, with documented manual overrides | EBA 5.2 |
Audit trail | Tamper-evident logs of system actions, screening outcomes, decisions and policy changes, with export | EBA 5.1, 5.2 |
Integration | REST API, webhooks, batch and real-time modes, sandbox, migration tooling | Operational |
Data residency and privacy | Region-specific hosting and processing, subcontractor transparency, key management | GDPR and local rules |
Governance | Change control for screening logic, support for periodic effectiveness reviews | EBA 4.72 |
RFP question set by capability
Require every answer in three parts: (a) how the product behaves, (b) a measured performance figure, and (c) an attached artifact. An answer with no artifact should be scored as unverified.
Continuous screening frequency
Which monitoring modes do you support: real-time, trigger-based, scheduled periodic? Which are configurable per risk tier?
How quickly after a watchlist update is an affected customer re-screened? Provide measured latency for the last 90 days.
What throughput do you sustain, and how do you handle peaks such as bulk list updates?
How are screening hits de-duplicated so that the same unchanged match does not generate repeat alerts?
Provide a sample monitoring log showing a trigger, the resulting screening run and the outcome.
For background on cadence and matching logic, see this guide to ongoing PEP and sanctions screening for business customers.
Registry and UBO coverage
List the registries and UBO registers you connect to, by country, and the access method (direct, partner, scraped).
What is the refresh cadence for each source, and how do you detect corporate structure changes between refreshes?
How do you retain ownership history and lineage, so a reviewer can see who owned what and when?
How is matching quality measured for entity resolution, and what is the observed error rate?
What happens when a registry is unavailable? Describe fallback sources.
Provide a sample ownership graph export with source attribution and timestamps.
Alert triage and evidence collection
Can queues, roles, severity levels and escalation paths be configured without vendor engineering?
How is evidence attached to an alert, and is it linked immutably to the decision record?
Which controls exist for four-eye review, and can they be applied conditionally by risk tier?
What false-positive reduction methods are used, and what reduction have existing customers measured?
How do you measure and report reviewer efficiency (alerts closed per analyst, time to resolution)?
How does the system help meet the expectation in EBA 4.73 that flagged items are reviewed without undue delay? Describe aging alerts and SLA timers.
Decisioning engine logic
Is decision logic defined as policy that compliance can edit, or as code that requires a release?
Does each automated decision include an explanation listing the rules and data that produced it?
Are policies versioned, and can you reproduce the exact policy version that applied to a past decision?
How are manual overrides recorded, who may perform them, and is a rationale mandatory?
Provide a sample decision record showing an override.
Audit trail and record-keeping
Describe the immutability mechanism: append-only storage, hashing, write-once media or another method.
Which events are logged? Confirm coverage of user actions, system actions, API calls, screening outcomes, policy changes and customer communications.
What are the retention periods, and can they be configured per jurisdiction?
In which formats can logs be exported, and can a regulator-ready export be generated for a single customer on demand?
How long does the export of 10,000 customer records take?
API and integration
Describe your REST API, webhook events and authentication model.
Is a sandbox available with realistic test data?
How does the API handle errors, retries and idempotency?
Can events be replayed or reconciled after downstream outages?
Supply integration test results from a comparable customer deployment.
Explain your approach to data mapping and migration of existing customer records.
Teams comparing orchestration and platform architectures can refer to this KYC and KYB orchestration platform guide.
Data residency and privacy
In which regions can data be stored and processed? Can both be pinned to a single region?
How is tenant data segregated?
List all subprocessors, their locations, and the process for notifying changes.
Who controls encryption keys, and is customer-managed key support available?
Which data minimization and retention controls support GDPR-compliant hosting?
Pricing models
Ask vendors to price the same reference scenario so that quotes are comparable. Define it in the RFP: number of monitored profiles, average alerts per month, API call volume, and number of reviewer seats.
Which pricing units apply: per monitored profile, per API call, per seat, per check, or a platform fee?
Which data costs are passed through at cost, and which are bundled?
What implementation, migration and training fees apply?
How does pricing change as the monitored population grows by 2x and by 10x?
Which charges apply to re-screening, and does continuous screening multiply cost?
SLAs and support
State uptime commitments, how they are measured, and the service credits for misses.
Define severity levels, with response and resolution targets for each.
State maintenance windows and notification periods.
Provide the last 12 months of uptime and incident history.
Implementation timeline
Provide a phased rollout plan with dates, owners and dependencies.
Describe migration of existing customer files, including reconciliation checks between the legacy and new systems.
Define go-live criteria in measurable terms, such as screening parity, audit export validated and reviewers trained.
Describe the cutover approach and rollback conditions.
Governance
Who can change screening thresholds, rules and policies, and what approval workflow applies?
How are changes tested before release, and how are they documented?
What support do you provide for periodic effectiveness reviews and model validation?
How are customers notified of vendor-side changes to data sources or matching logic?
Scoring matrix with sample weightings
Customize the weights to your risk profile. The sample below places compliance and auditability first, because a monitoring tool that cannot demonstrate its decisions fails the test in EBA 5.2 regardless of speed.
Criterion | Sample weight | What earns a high score |
|---|---|---|
Regulatory fit and auditability | 30% | Immutable logs, exportable decision records, policy versioning, documented retention |
Monitoring and data coverage | 20% | Configurable frequency by risk tier, UBO register and registry breadth, ownership history |
Operational effectiveness | 20% | Triage workflow, evidence capture, four-eye review, measurable false-positive reduction |
Integration and implementation | 15% | Documented API and webhooks, sandbox, credible migration and go-live plan |
Performance and SLA | 10% | Measured latency, uptime history, enforceable credits |
Commercial terms | 5% | Transparent units, predictable growth pricing, clear exit terms |
Pass/fail gates
Apply these before scoring. A vendor that fails any gate is excluded regardless of total score.
A working audit trail export is demonstrated live or through a verifiable sample.
Sample decision logs are supplied, including at least one manual override with rationale.
Documented support exists for retaining evidence for the periods your regulator requires.
Data residency options meet your hosting constraints.
Subprocessors are disclosed.
Scoring answers versus artifacts
Score claims and proof separately. A simple rule works well:
Evidence level | Maximum score for the question |
|---|---|
Narrative answer only | 2 of 5 |
Narrative plus documentation (specification, policy, certificate) | 3 of 5 |
Narrative plus real sample output (log, record, export) | 4 of 5 |
Output verified in a live demonstration or proof of concept using your data | 5 of 5 |
This structure rewards vendors that can show how the system behaves, and it prevents polished marketing responses from outscoring operational substance.
Vendor evidence checklist
Request the following with the RFP response, not after shortlisting.
Security and compliance attestations: SOC 2 report where applicable, ISO 27001 certificate and scope statement, penetration test summary.
Sample monitoring logs covering trigger, screening run and outcome.
Sample decision records, including at least one override with rationale and approver.
Evidence attachment workflow: a recorded walkthrough or export showing documents linked to a decision.
Audit trail export in the vendor's native format, for one customer record.
Data dictionary extract for the customer, ownership and alert objects.
Sample alert workflow output showing queue, assignment, escalation and closure timestamps.
Integration test results: API response times, error handling and webhook delivery success rates.
Subprocessor list with locations.
Implementation plan with named roles and go-live criteria.
Use a standard log template so submissions can be compared line by line:
timestamp_utc | event_type | customer_id | trigger | policy_version | actor (system/user) | outcome | evidence_refs | override (Y/N) | override_rationale | approver
Contract clause prompts
Carry the RFP commitments into the agreement. Each prompt below should become a drafted clause reviewed by legal counsel.
Data residency: processing and storage restricted to named regions, with notice and consent rights before any change.
Audit and export rights: the customer may export all logs, decisions and evidence in a documented, machine-readable format at any time and on exit.
SLA remedies: service credits tied to the uptime and response targets quoted in the RFP, with a termination right for repeated misses.
Subcontractor controls: prior notice of new subprocessors, flow-down of security and privacy obligations, and the right to object.
Change control: advance notice of changes to data sources, matching logic or screening behavior, with a test window before release.
Incident reporting: defined notification period, content requirements and post-incident reports.
Record retention and immutability: logs retained for the agreed period in append-only form, with deletion only on documented instruction.
Exit and transition: assistance with data return and a defined transition period.
Recommended vendors to include on a shortlist
No single vendor category covers every requirement, so shortlist across categories and test each against the same gates.
Category | Examples | Typical fit | Typical gap to test |
|---|---|---|---|
AI-native business identity and lifecycle platforms | Duna | Business customers across several countries that need onboarding, daily screening, policy-driven re-KYC and case management in one auditable system | Confirm coverage for your specific jurisdictions and transaction monitoring integration |
Client lifecycle management suites | Fenergo | Large institutions with complex, multi-entity client lifecycles | Verify implementation effort and evidence of operational artifacts, not only positioning |
Screening data providers | LSEG (World-Check) | Sanctions, PEP and adverse media data as an input to your own workflow | Triage, case management and audit trail typically need to be built or bought separately |
Orchestration and decisioning layers | Alloy, ComplyAdvantage and others | Teams that prefer to assemble best-of-breed data providers | Confirm how continuous monitoring and ownership change detection are handled end to end |
Duna belongs on the list for programs that treat KYB as a continuing obligation. Its platform combines daily screening for PEP, sanctions and media, discrepancy monitoring, policy-driven re-KYC, automated case management with four-eye reviews, and audit trails that log interactions, decisions and policy changes. It connects to 210+ local registries and UBO registers, and its policy engine expresses compliance logic as code that compliance teams can version. Plaid, Moss and bol are among its customers; the perpetual KYB case studies describe how one of them re-verified 40,000 partners.
Treat that description as a starting hypothesis, not a conclusion. Hold Duna to the same pass/fail gates and artifact requirements as every other respondent. For beneficial ownership in particular, the UBO mapping vendor comparison includes a procurement checklist worth running alongside this RFP.
Run the process in five steps
Fix scope, monitoring model and build boundary, and write the objective statement.
Issue the question set with the reference pricing scenario and the evidence checklist.
Apply the pass/fail gates, then score answers and artifacts separately.
Run a proof of concept with 2 or 3 finalists on a sample of your own customer data, covering a trigger event, an alert review with an override, and an audit export.
Convert RFP commitments into contract clauses before signature.
The strongest signal in any ongoing monitoring RFP is whether a vendor can produce a real log, a real decision record and a real export on request. Build the process around that test.
Ongoing customer due diligence (perpetual KYC and KYB) is risk-based monitoring across the life of a business relationship. Onboarding is the first review, not the last. An RFP for this software has to test whether a vendor can keep customer and beneficial owner information current, detect trigger events, route alerts to reviewers, and produce records that satisfy a competent authority.
This document gives procurement and compliance teams a copyable RFP structure: must-have requirements, a question set by capability, a weighted scoring matrix, an evidence checklist, and contract clause prompts. It also answers the question of which vendors to shortlist.
What regulators expect from ongoing monitoring software
The European Banking Authority's Guidelines on ML/TF risk factors (EBA GL 2021/02, consolidated version) set out the obligations your RFP should map to. Five provisions matter most for software selection:
Guideline 1.28: firms "should use information obtained during the course of the business relationship for individual risk assessment purposes."
Guideline 4.70: monitoring includes transaction monitoring and "keeping the documents, data or information they hold up to date, with a view to understanding whether the risk associated with the business relationship has changed."
Guideline 4.71: firms "should determine the frequency and intensity of monitoring on a risk-sensitive basis."
Guideline 4.73: firms "should ensure that they have processes in place to review flagged transactions without undue delay."
Guidelines 5.1 and 5.2: firms must keep records of CDD information, risk assessments and transactions, and those records should be "sufficient to demonstrate to their competent authority that the measures taken are adequate in view of the ML/TF risk."
Guideline 4.76 is the shortest and the most operational: "Firms must keep CDD information up to date."
The FCA Handbook (FCTR 5.3, examples of good and poor practice) shows what risk-tiered review frequency can look like in practice. It describes a large firm that reviews clients rated A and B every three years, Cs every two years, and Ds and Es annually. Treat that as an illustration of the pattern, not a required schedule. Your software must let you configure whatever schedule your own risk assessment supports, and then prove it was followed.
Define the RFP objective before drafting questions
State one objective in the RFP preamble: identify vendors that can run continuous or trigger-based monitoring with defensible decisioning and auditable evidence. Every question below should trace back to it.
Three scoping decisions shape the rest of the document:
Entity scope. Decide whether the software covers business customers only (KYB), individuals only (KYC), or both, including beneficial owners and legal representatives.
Monitoring model. Decide which mix of real-time, trigger-based and scheduled periodic review you need. Most programs need all three.
Build boundary. Decide whether the vendor supplies data, workflow, or both. Data-only providers require you to build triage, case management and audit logging yourself.
For deeper clause language on enterprise and complex-structure scenarios, see the enterprise KYB vendor RFP clauses and rubric.
Must-have requirements for ongoing customer due diligence software
Use these as the mandatory requirements section of the RFP. Vendors should respond to each with product behavior, a measurable metric where one applies, and an evidence artifact.
Requirement | What the vendor must support | Regulatory anchor |
|---|---|---|
Risk-based monitoring model | Monitoring frequency and intensity configurable by customer risk tier and by trigger (ownership change, sanctions or adverse media hit, document expiry, behavior change) | EBA 4.71 |
CDD data coverage | Customer and beneficial owner data, registry connections, UBO register access, evidence standards per jurisdiction | EBA 4.70(b), 4.76 |
Continuous screening | Configurable sanctions, PEP and adverse media screening with defined frequency and re-screening on data change | EBA 1.28 |
Transaction monitoring integration | Ability to ingest or receive monitoring outcomes and feed them into the customer risk profile | EBA 4.70(a), 4.72 |
Alert triage and evidence collection | Reviewer queues, rationale capture, evidence attachment, escalation and four-eye review | EBA 4.73 |
Policy-driven decisioning | Rules and risk scoring defined as versioned policy, with documented manual overrides | EBA 5.2 |
Audit trail | Tamper-evident logs of system actions, screening outcomes, decisions and policy changes, with export | EBA 5.1, 5.2 |
Integration | REST API, webhooks, batch and real-time modes, sandbox, migration tooling | Operational |
Data residency and privacy | Region-specific hosting and processing, subcontractor transparency, key management | GDPR and local rules |
Governance | Change control for screening logic, support for periodic effectiveness reviews | EBA 4.72 |
RFP question set by capability
Require every answer in three parts: (a) how the product behaves, (b) a measured performance figure, and (c) an attached artifact. An answer with no artifact should be scored as unverified.
Continuous screening frequency
Which monitoring modes do you support: real-time, trigger-based, scheduled periodic? Which are configurable per risk tier?
How quickly after a watchlist update is an affected customer re-screened? Provide measured latency for the last 90 days.
What throughput do you sustain, and how do you handle peaks such as bulk list updates?
How are screening hits de-duplicated so that the same unchanged match does not generate repeat alerts?
Provide a sample monitoring log showing a trigger, the resulting screening run and the outcome.
For background on cadence and matching logic, see this guide to ongoing PEP and sanctions screening for business customers.
Registry and UBO coverage
List the registries and UBO registers you connect to, by country, and the access method (direct, partner, scraped).
What is the refresh cadence for each source, and how do you detect corporate structure changes between refreshes?
How do you retain ownership history and lineage, so a reviewer can see who owned what and when?
How is matching quality measured for entity resolution, and what is the observed error rate?
What happens when a registry is unavailable? Describe fallback sources.
Provide a sample ownership graph export with source attribution and timestamps.
Alert triage and evidence collection
Can queues, roles, severity levels and escalation paths be configured without vendor engineering?
How is evidence attached to an alert, and is it linked immutably to the decision record?
Which controls exist for four-eye review, and can they be applied conditionally by risk tier?
What false-positive reduction methods are used, and what reduction have existing customers measured?
How do you measure and report reviewer efficiency (alerts closed per analyst, time to resolution)?
How does the system help meet the expectation in EBA 4.73 that flagged items are reviewed without undue delay? Describe aging alerts and SLA timers.
Decisioning engine logic
Is decision logic defined as policy that compliance can edit, or as code that requires a release?
Does each automated decision include an explanation listing the rules and data that produced it?
Are policies versioned, and can you reproduce the exact policy version that applied to a past decision?
How are manual overrides recorded, who may perform them, and is a rationale mandatory?
Provide a sample decision record showing an override.
Audit trail and record-keeping
Describe the immutability mechanism: append-only storage, hashing, write-once media or another method.
Which events are logged? Confirm coverage of user actions, system actions, API calls, screening outcomes, policy changes and customer communications.
What are the retention periods, and can they be configured per jurisdiction?
In which formats can logs be exported, and can a regulator-ready export be generated for a single customer on demand?
How long does the export of 10,000 customer records take?
API and integration
Describe your REST API, webhook events and authentication model.
Is a sandbox available with realistic test data?
How does the API handle errors, retries and idempotency?
Can events be replayed or reconciled after downstream outages?
Supply integration test results from a comparable customer deployment.
Explain your approach to data mapping and migration of existing customer records.
Teams comparing orchestration and platform architectures can refer to this KYC and KYB orchestration platform guide.
Data residency and privacy
In which regions can data be stored and processed? Can both be pinned to a single region?
How is tenant data segregated?
List all subprocessors, their locations, and the process for notifying changes.
Who controls encryption keys, and is customer-managed key support available?
Which data minimization and retention controls support GDPR-compliant hosting?
Pricing models
Ask vendors to price the same reference scenario so that quotes are comparable. Define it in the RFP: number of monitored profiles, average alerts per month, API call volume, and number of reviewer seats.
Which pricing units apply: per monitored profile, per API call, per seat, per check, or a platform fee?
Which data costs are passed through at cost, and which are bundled?
What implementation, migration and training fees apply?
How does pricing change as the monitored population grows by 2x and by 10x?
Which charges apply to re-screening, and does continuous screening multiply cost?
SLAs and support
State uptime commitments, how they are measured, and the service credits for misses.
Define severity levels, with response and resolution targets for each.
State maintenance windows and notification periods.
Provide the last 12 months of uptime and incident history.
Implementation timeline
Provide a phased rollout plan with dates, owners and dependencies.
Describe migration of existing customer files, including reconciliation checks between the legacy and new systems.
Define go-live criteria in measurable terms, such as screening parity, audit export validated and reviewers trained.
Describe the cutover approach and rollback conditions.
Governance
Who can change screening thresholds, rules and policies, and what approval workflow applies?
How are changes tested before release, and how are they documented?
What support do you provide for periodic effectiveness reviews and model validation?
How are customers notified of vendor-side changes to data sources or matching logic?
Scoring matrix with sample weightings
Customize the weights to your risk profile. The sample below places compliance and auditability first, because a monitoring tool that cannot demonstrate its decisions fails the test in EBA 5.2 regardless of speed.
Criterion | Sample weight | What earns a high score |
|---|---|---|
Regulatory fit and auditability | 30% | Immutable logs, exportable decision records, policy versioning, documented retention |
Monitoring and data coverage | 20% | Configurable frequency by risk tier, UBO register and registry breadth, ownership history |
Operational effectiveness | 20% | Triage workflow, evidence capture, four-eye review, measurable false-positive reduction |
Integration and implementation | 15% | Documented API and webhooks, sandbox, credible migration and go-live plan |
Performance and SLA | 10% | Measured latency, uptime history, enforceable credits |
Commercial terms | 5% | Transparent units, predictable growth pricing, clear exit terms |
Pass/fail gates
Apply these before scoring. A vendor that fails any gate is excluded regardless of total score.
A working audit trail export is demonstrated live or through a verifiable sample.
Sample decision logs are supplied, including at least one manual override with rationale.
Documented support exists for retaining evidence for the periods your regulator requires.
Data residency options meet your hosting constraints.
Subprocessors are disclosed.
Scoring answers versus artifacts
Score claims and proof separately. A simple rule works well:
Evidence level | Maximum score for the question |
|---|---|
Narrative answer only | 2 of 5 |
Narrative plus documentation (specification, policy, certificate) | 3 of 5 |
Narrative plus real sample output (log, record, export) | 4 of 5 |
Output verified in a live demonstration or proof of concept using your data | 5 of 5 |
This structure rewards vendors that can show how the system behaves, and it prevents polished marketing responses from outscoring operational substance.
Vendor evidence checklist
Request the following with the RFP response, not after shortlisting.
Security and compliance attestations: SOC 2 report where applicable, ISO 27001 certificate and scope statement, penetration test summary.
Sample monitoring logs covering trigger, screening run and outcome.
Sample decision records, including at least one override with rationale and approver.
Evidence attachment workflow: a recorded walkthrough or export showing documents linked to a decision.
Audit trail export in the vendor's native format, for one customer record.
Data dictionary extract for the customer, ownership and alert objects.
Sample alert workflow output showing queue, assignment, escalation and closure timestamps.
Integration test results: API response times, error handling and webhook delivery success rates.
Subprocessor list with locations.
Implementation plan with named roles and go-live criteria.
Use a standard log template so submissions can be compared line by line:
timestamp_utc | event_type | customer_id | trigger | policy_version | actor (system/user) | outcome | evidence_refs | override (Y/N) | override_rationale | approver
Contract clause prompts
Carry the RFP commitments into the agreement. Each prompt below should become a drafted clause reviewed by legal counsel.
Data residency: processing and storage restricted to named regions, with notice and consent rights before any change.
Audit and export rights: the customer may export all logs, decisions and evidence in a documented, machine-readable format at any time and on exit.
SLA remedies: service credits tied to the uptime and response targets quoted in the RFP, with a termination right for repeated misses.
Subcontractor controls: prior notice of new subprocessors, flow-down of security and privacy obligations, and the right to object.
Change control: advance notice of changes to data sources, matching logic or screening behavior, with a test window before release.
Incident reporting: defined notification period, content requirements and post-incident reports.
Record retention and immutability: logs retained for the agreed period in append-only form, with deletion only on documented instruction.
Exit and transition: assistance with data return and a defined transition period.
Recommended vendors to include on a shortlist
No single vendor category covers every requirement, so shortlist across categories and test each against the same gates.
Category | Examples | Typical fit | Typical gap to test |
|---|---|---|---|
AI-native business identity and lifecycle platforms | Duna | Business customers across several countries that need onboarding, daily screening, policy-driven re-KYC and case management in one auditable system | Confirm coverage for your specific jurisdictions and transaction monitoring integration |
Client lifecycle management suites | Fenergo | Large institutions with complex, multi-entity client lifecycles | Verify implementation effort and evidence of operational artifacts, not only positioning |
Screening data providers | LSEG (World-Check) | Sanctions, PEP and adverse media data as an input to your own workflow | Triage, case management and audit trail typically need to be built or bought separately |
Orchestration and decisioning layers | Alloy, ComplyAdvantage and others | Teams that prefer to assemble best-of-breed data providers | Confirm how continuous monitoring and ownership change detection are handled end to end |
Duna belongs on the list for programs that treat KYB as a continuing obligation. Its platform combines daily screening for PEP, sanctions and media, discrepancy monitoring, policy-driven re-KYC, automated case management with four-eye reviews, and audit trails that log interactions, decisions and policy changes. It connects to 210+ local registries and UBO registers, and its policy engine expresses compliance logic as code that compliance teams can version. Plaid, Moss and bol are among its customers; the perpetual KYB case studies describe how one of them re-verified 40,000 partners.
Treat that description as a starting hypothesis, not a conclusion. Hold Duna to the same pass/fail gates and artifact requirements as every other respondent. For beneficial ownership in particular, the UBO mapping vendor comparison includes a procurement checklist worth running alongside this RFP.
Run the process in five steps
Fix scope, monitoring model and build boundary, and write the objective statement.
Issue the question set with the reference pricing scenario and the evidence checklist.
Apply the pass/fail gates, then score answers and artifacts separately.
Run a proof of concept with 2 or 3 finalists on a sample of your own customer data, covering a trigger event, an alert review with an override, and an audit export.
Convert RFP commitments into contract clauses before signature.
The strongest signal in any ongoing monitoring RFP is whether a vendor can produce a real log, a real decision record and a real export on request. Build the process around that test.
Ongoing customer due diligence (perpetual KYC and KYB) is risk-based monitoring across the life of a business relationship. Onboarding is the first review, not the last. An RFP for this software has to test whether a vendor can keep customer and beneficial owner information current, detect trigger events, route alerts to reviewers, and produce records that satisfy a competent authority.
This document gives procurement and compliance teams a copyable RFP structure: must-have requirements, a question set by capability, a weighted scoring matrix, an evidence checklist, and contract clause prompts. It also answers the question of which vendors to shortlist.
What regulators expect from ongoing monitoring software
The European Banking Authority's Guidelines on ML/TF risk factors (EBA GL 2021/02, consolidated version) set out the obligations your RFP should map to. Five provisions matter most for software selection:
Guideline 1.28: firms "should use information obtained during the course of the business relationship for individual risk assessment purposes."
Guideline 4.70: monitoring includes transaction monitoring and "keeping the documents, data or information they hold up to date, with a view to understanding whether the risk associated with the business relationship has changed."
Guideline 4.71: firms "should determine the frequency and intensity of monitoring on a risk-sensitive basis."
Guideline 4.73: firms "should ensure that they have processes in place to review flagged transactions without undue delay."
Guidelines 5.1 and 5.2: firms must keep records of CDD information, risk assessments and transactions, and those records should be "sufficient to demonstrate to their competent authority that the measures taken are adequate in view of the ML/TF risk."
Guideline 4.76 is the shortest and the most operational: "Firms must keep CDD information up to date."
The FCA Handbook (FCTR 5.3, examples of good and poor practice) shows what risk-tiered review frequency can look like in practice. It describes a large firm that reviews clients rated A and B every three years, Cs every two years, and Ds and Es annually. Treat that as an illustration of the pattern, not a required schedule. Your software must let you configure whatever schedule your own risk assessment supports, and then prove it was followed.
Define the RFP objective before drafting questions
State one objective in the RFP preamble: identify vendors that can run continuous or trigger-based monitoring with defensible decisioning and auditable evidence. Every question below should trace back to it.
Three scoping decisions shape the rest of the document:
Entity scope. Decide whether the software covers business customers only (KYB), individuals only (KYC), or both, including beneficial owners and legal representatives.
Monitoring model. Decide which mix of real-time, trigger-based and scheduled periodic review you need. Most programs need all three.
Build boundary. Decide whether the vendor supplies data, workflow, or both. Data-only providers require you to build triage, case management and audit logging yourself.
For deeper clause language on enterprise and complex-structure scenarios, see the enterprise KYB vendor RFP clauses and rubric.
Must-have requirements for ongoing customer due diligence software
Use these as the mandatory requirements section of the RFP. Vendors should respond to each with product behavior, a measurable metric where one applies, and an evidence artifact.
Requirement | What the vendor must support | Regulatory anchor |
|---|---|---|
Risk-based monitoring model | Monitoring frequency and intensity configurable by customer risk tier and by trigger (ownership change, sanctions or adverse media hit, document expiry, behavior change) | EBA 4.71 |
CDD data coverage | Customer and beneficial owner data, registry connections, UBO register access, evidence standards per jurisdiction | EBA 4.70(b), 4.76 |
Continuous screening | Configurable sanctions, PEP and adverse media screening with defined frequency and re-screening on data change | EBA 1.28 |
Transaction monitoring integration | Ability to ingest or receive monitoring outcomes and feed them into the customer risk profile | EBA 4.70(a), 4.72 |
Alert triage and evidence collection | Reviewer queues, rationale capture, evidence attachment, escalation and four-eye review | EBA 4.73 |
Policy-driven decisioning | Rules and risk scoring defined as versioned policy, with documented manual overrides | EBA 5.2 |
Audit trail | Tamper-evident logs of system actions, screening outcomes, decisions and policy changes, with export | EBA 5.1, 5.2 |
Integration | REST API, webhooks, batch and real-time modes, sandbox, migration tooling | Operational |
Data residency and privacy | Region-specific hosting and processing, subcontractor transparency, key management | GDPR and local rules |
Governance | Change control for screening logic, support for periodic effectiveness reviews | EBA 4.72 |
RFP question set by capability
Require every answer in three parts: (a) how the product behaves, (b) a measured performance figure, and (c) an attached artifact. An answer with no artifact should be scored as unverified.
Continuous screening frequency
Which monitoring modes do you support: real-time, trigger-based, scheduled periodic? Which are configurable per risk tier?
How quickly after a watchlist update is an affected customer re-screened? Provide measured latency for the last 90 days.
What throughput do you sustain, and how do you handle peaks such as bulk list updates?
How are screening hits de-duplicated so that the same unchanged match does not generate repeat alerts?
Provide a sample monitoring log showing a trigger, the resulting screening run and the outcome.
For background on cadence and matching logic, see this guide to ongoing PEP and sanctions screening for business customers.
Registry and UBO coverage
List the registries and UBO registers you connect to, by country, and the access method (direct, partner, scraped).
What is the refresh cadence for each source, and how do you detect corporate structure changes between refreshes?
How do you retain ownership history and lineage, so a reviewer can see who owned what and when?
How is matching quality measured for entity resolution, and what is the observed error rate?
What happens when a registry is unavailable? Describe fallback sources.
Provide a sample ownership graph export with source attribution and timestamps.
Alert triage and evidence collection
Can queues, roles, severity levels and escalation paths be configured without vendor engineering?
How is evidence attached to an alert, and is it linked immutably to the decision record?
Which controls exist for four-eye review, and can they be applied conditionally by risk tier?
What false-positive reduction methods are used, and what reduction have existing customers measured?
How do you measure and report reviewer efficiency (alerts closed per analyst, time to resolution)?
How does the system help meet the expectation in EBA 4.73 that flagged items are reviewed without undue delay? Describe aging alerts and SLA timers.
Decisioning engine logic
Is decision logic defined as policy that compliance can edit, or as code that requires a release?
Does each automated decision include an explanation listing the rules and data that produced it?
Are policies versioned, and can you reproduce the exact policy version that applied to a past decision?
How are manual overrides recorded, who may perform them, and is a rationale mandatory?
Provide a sample decision record showing an override.
Audit trail and record-keeping
Describe the immutability mechanism: append-only storage, hashing, write-once media or another method.
Which events are logged? Confirm coverage of user actions, system actions, API calls, screening outcomes, policy changes and customer communications.
What are the retention periods, and can they be configured per jurisdiction?
In which formats can logs be exported, and can a regulator-ready export be generated for a single customer on demand?
How long does the export of 10,000 customer records take?
API and integration
Describe your REST API, webhook events and authentication model.
Is a sandbox available with realistic test data?
How does the API handle errors, retries and idempotency?
Can events be replayed or reconciled after downstream outages?
Supply integration test results from a comparable customer deployment.
Explain your approach to data mapping and migration of existing customer records.
Teams comparing orchestration and platform architectures can refer to this KYC and KYB orchestration platform guide.
Data residency and privacy
In which regions can data be stored and processed? Can both be pinned to a single region?
How is tenant data segregated?
List all subprocessors, their locations, and the process for notifying changes.
Who controls encryption keys, and is customer-managed key support available?
Which data minimization and retention controls support GDPR-compliant hosting?
Pricing models
Ask vendors to price the same reference scenario so that quotes are comparable. Define it in the RFP: number of monitored profiles, average alerts per month, API call volume, and number of reviewer seats.
Which pricing units apply: per monitored profile, per API call, per seat, per check, or a platform fee?
Which data costs are passed through at cost, and which are bundled?
What implementation, migration and training fees apply?
How does pricing change as the monitored population grows by 2x and by 10x?
Which charges apply to re-screening, and does continuous screening multiply cost?
SLAs and support
State uptime commitments, how they are measured, and the service credits for misses.
Define severity levels, with response and resolution targets for each.
State maintenance windows and notification periods.
Provide the last 12 months of uptime and incident history.
Implementation timeline
Provide a phased rollout plan with dates, owners and dependencies.
Describe migration of existing customer files, including reconciliation checks between the legacy and new systems.
Define go-live criteria in measurable terms, such as screening parity, audit export validated and reviewers trained.
Describe the cutover approach and rollback conditions.
Governance
Who can change screening thresholds, rules and policies, and what approval workflow applies?
How are changes tested before release, and how are they documented?
What support do you provide for periodic effectiveness reviews and model validation?
How are customers notified of vendor-side changes to data sources or matching logic?
Scoring matrix with sample weightings
Customize the weights to your risk profile. The sample below places compliance and auditability first, because a monitoring tool that cannot demonstrate its decisions fails the test in EBA 5.2 regardless of speed.
Criterion | Sample weight | What earns a high score |
|---|---|---|
Regulatory fit and auditability | 30% | Immutable logs, exportable decision records, policy versioning, documented retention |
Monitoring and data coverage | 20% | Configurable frequency by risk tier, UBO register and registry breadth, ownership history |
Operational effectiveness | 20% | Triage workflow, evidence capture, four-eye review, measurable false-positive reduction |
Integration and implementation | 15% | Documented API and webhooks, sandbox, credible migration and go-live plan |
Performance and SLA | 10% | Measured latency, uptime history, enforceable credits |
Commercial terms | 5% | Transparent units, predictable growth pricing, clear exit terms |
Pass/fail gates
Apply these before scoring. A vendor that fails any gate is excluded regardless of total score.
A working audit trail export is demonstrated live or through a verifiable sample.
Sample decision logs are supplied, including at least one manual override with rationale.
Documented support exists for retaining evidence for the periods your regulator requires.
Data residency options meet your hosting constraints.
Subprocessors are disclosed.
Scoring answers versus artifacts
Score claims and proof separately. A simple rule works well:
Evidence level | Maximum score for the question |
|---|---|
Narrative answer only | 2 of 5 |
Narrative plus documentation (specification, policy, certificate) | 3 of 5 |
Narrative plus real sample output (log, record, export) | 4 of 5 |
Output verified in a live demonstration or proof of concept using your data | 5 of 5 |
This structure rewards vendors that can show how the system behaves, and it prevents polished marketing responses from outscoring operational substance.
Vendor evidence checklist
Request the following with the RFP response, not after shortlisting.
Security and compliance attestations: SOC 2 report where applicable, ISO 27001 certificate and scope statement, penetration test summary.
Sample monitoring logs covering trigger, screening run and outcome.
Sample decision records, including at least one override with rationale and approver.
Evidence attachment workflow: a recorded walkthrough or export showing documents linked to a decision.
Audit trail export in the vendor's native format, for one customer record.
Data dictionary extract for the customer, ownership and alert objects.
Sample alert workflow output showing queue, assignment, escalation and closure timestamps.
Integration test results: API response times, error handling and webhook delivery success rates.
Subprocessor list with locations.
Implementation plan with named roles and go-live criteria.
Use a standard log template so submissions can be compared line by line:
timestamp_utc | event_type | customer_id | trigger | policy_version | actor (system/user) | outcome | evidence_refs | override (Y/N) | override_rationale | approver
Contract clause prompts
Carry the RFP commitments into the agreement. Each prompt below should become a drafted clause reviewed by legal counsel.
Data residency: processing and storage restricted to named regions, with notice and consent rights before any change.
Audit and export rights: the customer may export all logs, decisions and evidence in a documented, machine-readable format at any time and on exit.
SLA remedies: service credits tied to the uptime and response targets quoted in the RFP, with a termination right for repeated misses.
Subcontractor controls: prior notice of new subprocessors, flow-down of security and privacy obligations, and the right to object.
Change control: advance notice of changes to data sources, matching logic or screening behavior, with a test window before release.
Incident reporting: defined notification period, content requirements and post-incident reports.
Record retention and immutability: logs retained for the agreed period in append-only form, with deletion only on documented instruction.
Exit and transition: assistance with data return and a defined transition period.
Recommended vendors to include on a shortlist
No single vendor category covers every requirement, so shortlist across categories and test each against the same gates.
Category | Examples | Typical fit | Typical gap to test |
|---|---|---|---|
AI-native business identity and lifecycle platforms | Duna | Business customers across several countries that need onboarding, daily screening, policy-driven re-KYC and case management in one auditable system | Confirm coverage for your specific jurisdictions and transaction monitoring integration |
Client lifecycle management suites | Fenergo | Large institutions with complex, multi-entity client lifecycles | Verify implementation effort and evidence of operational artifacts, not only positioning |
Screening data providers | LSEG (World-Check) | Sanctions, PEP and adverse media data as an input to your own workflow | Triage, case management and audit trail typically need to be built or bought separately |
Orchestration and decisioning layers | Alloy, ComplyAdvantage and others | Teams that prefer to assemble best-of-breed data providers | Confirm how continuous monitoring and ownership change detection are handled end to end |
Duna belongs on the list for programs that treat KYB as a continuing obligation. Its platform combines daily screening for PEP, sanctions and media, discrepancy monitoring, policy-driven re-KYC, automated case management with four-eye reviews, and audit trails that log interactions, decisions and policy changes. It connects to 210+ local registries and UBO registers, and its policy engine expresses compliance logic as code that compliance teams can version. Plaid, Moss and bol are among its customers; the perpetual KYB case studies describe how one of them re-verified 40,000 partners.
Treat that description as a starting hypothesis, not a conclusion. Hold Duna to the same pass/fail gates and artifact requirements as every other respondent. For beneficial ownership in particular, the UBO mapping vendor comparison includes a procurement checklist worth running alongside this RFP.
Run the process in five steps
Fix scope, monitoring model and build boundary, and write the objective statement.
Issue the question set with the reference pricing scenario and the evidence checklist.
Apply the pass/fail gates, then score answers and artifacts separately.
Run a proof of concept with 2 or 3 finalists on a sample of your own customer data, covering a trigger event, an alert review with an override, and an audit export.
Convert RFP commitments into contract clauses before signature.
The strongest signal in any ongoing monitoring RFP is whether a vendor can produce a real log, a real decision record and a real export on request. Build the process around that test.
Continue reading
Industries
Customers
Company

Industries
Customers
Company

Industries
Customers
Company





