Baobyte AML Suite
Book a demo

Commercial banks

When the hard part is not detection, but agreeing on the answer

Tier-1 and Tier-2 banks rarely lack monitoring rules. What they lack is one place where a decision, the policy version behind it and the evidence that supported it stay together across branches, subsidiaries and a compliance function of dozens of people.

  • CBN-licensed
  • Federated compliance teams
  • Branch and subsidiary org units
  • High transaction volume

Obligations

What a licensed bank has to be able to demonstrate

Stated qualitatively on purpose. Thresholds, filing windows and instrument references change, and your compliance counsel — not a vendor's marketing page — should be the source for the current figure.

CBNA board-approved programme with named accountable officers
A licensed bank is expected to maintain a documented, board-approved AML/CFT/CPF compliance programme, appoint senior compliance and money-laundering reporting officers with sufficient authority and independence, and subject the programme to independent testing. The programme has to be evidenced, not asserted: policy versions, approval records, training completion and the results of the independent review.
CBNEnterprise-wide risk assessment that drives the controls
Risk assessment across customers, products, delivery channels and geographies, refreshed on a defined cycle, with the resulting risk appetite visibly connected to the CDD tiers, monitoring thresholds and approval levels actually configured in the system. Supervisors increasingly test that link rather than the document.
NFIUSuspicious and threshold transaction reporting
Suspicious transactions are reported to the Nigerian Financial Intelligence Unit, and currency and threshold transactions are rendered on the prescribed cycle. The decision not to report is examined as closely as the decision to report, so the reasoning behind a closed alert has to survive being read years later by someone who was not there.
FATF-alignedGroup-wide programme across branches and subsidiaries
A banking group is expected to apply its AML/CFT programme across branches and majority-owned subsidiaries, including sharing information for risk-management purposes, subject to host-country law. In practice that means the group has to be able to see the same customer the same way in two entities running two different core systems.
FATF-alignedCorrespondent banking and cross-border transfers
Cross-border correspondent relationships attract enhanced due diligence and senior approval, shell-bank relationships are prohibited, and originator and beneficiary information is expected to travel with a transfer and to be usable for screening on receipt. Incoming transfers with incomplete party information need a defined, evidenced handling policy rather than an informal one.
SanctionsTargeted financial sanctions applied without delay
Sanctions obligations are not risk-based: designated parties are to be identified and their assets frozen without delay, which makes list currency, screening coverage of the existing book and the speed of the escalation path the things that matter. The platform retains the list source and version behind every hit so a freeze decision can be tied to the list as it stood at that moment.
NDPALawful processing of the data the programme depends on
Customer identity documents, transaction histories and investigation narratives are personal data. Retention has to be defined and defensible, subject-rights requests answered without breaking the audit record, and access to sensitive fields restricted and reviewable.
Internal auditIndependent testing that can retrieve its own evidence
The third line is expected to test the programme without depending on the second line to assemble the sample. That is a systems requirement as much as a governance one — read-only, independently retrievable evidence rather than a folder someone prepared.

Operational reality

What scale actually costs you

These are the six things Tier-1 and Tier-2 compliance functions describe when asked what breaks — and none of them is a missing detection scenario.

  • Eleven people own one decisionA high-risk onboarding at a Tier-1 bank passes through the relationship manager, branch compliance, regional compliance, the sanctions desk, EDD, the MLRO and sometimes a committee. Each hand-off today is usually an email with an attachment. The decision is defensible only if the chain is reconstructable, and email is not a chain.
  • The same customer gets two answers in two branchesPolicy is uniform on paper and interpreted locally in practice. Without a shared risk model, shared forms and a shared case history, Lagos and Kano can rate the same profile differently and neither will know. The examiner will.
  • Tuning a threshold is a governance event, not a config changeLowering a rule threshold changes detection volume, alert backlog and — if it is ever questioned — the institution's story about what it was monitoring last year. That change needs a proposer, an approver, an effective date and the ability to replay the old version deterministically against the same data.
  • Alert volume outruns analyst capacity, and the queue hides the riskWhen the queue is thirty thousand deep, the constraint stops being detection and becomes triage. Routing, prioritisation, and the ability to see that four alerts belong to one subject rather than four are worth more than another rule.
  • Evidence lives in six systems and one examiner wants all of itCore banking, the card switch, the screening tool, the case tracker, a shared drive and someone's inbox. Assembling four years of history for one sampled customer is a multi-week exercise that produces a bundle nobody can prove is complete.
  • Thousands of users, and access that driftsJoiners, movers and leavers across hundreds of branches means permission drift is continuous. The question a supervisor asks is not whether roles exist but who could see this customer's identifiers last March, and whether anyone reviewed that.

Scale

The same control, at two different sizes

Every control below exists at a microfinance bank too. What changes is that at Tier-1 scale the informal version stops being auditable.

How a control changes between a small institution and a Tier-1 bank
ControlSmaller institutionTier-1 reality
Rule changeTwo people agree and someone edits the configProposal, approval, effective date, retained prior version and deterministic replay
Case assignmentWhoever is free picks it upRouting by org unit, product and risk, with the assignment history part of the case record
Identifier accessThe compliance team can see everythingReveal is a permissioned, audited event with an accountable approver
Board reportingA spreadsheet assembled the week beforeAggregate analytics that are suppression-aware and fail closed when a data stream is unavailable
Examination sampleOne analyst pulls the filesIndependent, read-only retrieval by the third line with chain verification output
Group viewNot applicableOrg-unit scoping within one tenant, so a subsidiary's staff see their own book and the group can still see the whole

Platform

Which capabilities carry this segment

Each links to the capability page. The sentence is what that capability does for a bank specifically.

Transaction monitoring

Explainable rules with retained version and evaluation evidence, deterministic replay of a past decision, and a tuning path that runs through proposal and approval rather than a console edit.

Learn more

Investigation and casework

A subject-consistent workspace that pulls the alerts, screening hits, transactions, CDD and documents for one subject into one case, with assignment history and four-eyes closure.

Learn more

Screening and watchlists

Sanctions, PEP and adverse-media adjudication that retains source, source version and match evidence, so a cleared hit can be defended and a reopened one has a reason attached.

Learn more

Customer and CDD

Governed CDD forms, ownership declarations that are cycle-safe and historical, and risk assessment configured centrally so branch interpretation cannot quietly diverge.

Learn more

Governed reporting

STR, CTR and institution returns generated from governed mappings, with review, approval and filing evidence held against the exact payload that was submitted.

Learn more

Audit and privacy

Append-only per-tenant audit chain, sensitive-access review, retention schedules and legal holds — the evidence the third line and the supervisor both ask for.

Learn more

Integrations

Core banking and switch ingest with idempotency, currency safety and quarantine of malformed input, so the monitoring layer is never quietly running on a partial feed.

Learn more

Security architecture

Tenant isolation enforced in the database, federated identity through your existing IdP, and separated human and machine identities — the section your security review will go to first.

Learn more

Examination readiness

What you can put in front of an examiner

The test of an AML platform at this scale is not what it detects. It is what it can produce, unassisted, about something it decided three years ago.

  • The rule, rule version and evaluation evidence behind any alert, including ones that were closed without escalation.
  • The full assignment and decision history of a case, with the approval that closed it and the actor who gave it.
  • The screening list source and version as they stood when a hit was raised, cleared, reopened or suppressed.
  • Every sensitive-identifier reveal, with actor, reason and timestamp, sampled and reviewable independently.
  • Configuration history: which risk model, which CDD form version and which monitoring thresholds were live on a given date, and who approved them.
  • Filing evidence tied to the exact payload digest submitted, with the acknowledgement state from the channel.

Proof

Commercial bank case study

This slot is reserved for a Challenge → Approach → Result write-up from a bank that has approved publication.

Placeholder — no commercial bank engagement has been approved for publication. Replace with sourced, approved content before this page is made public.

Reserved for a bank that has approved publication

When a commercial bank customer agrees to be written about, this becomes a structured case study. Until then it stays visibly empty rather than being filled with a plausible composite.

Challenge
The institution's starting position, in its own words and with its own numbers.
Approach
What was configured, what was integrated, and how long each phase actually took.
Result
Measured outcomes with a defined baseline and measurement period, signed off by the institution.

—

Alert backlog change

Awaiting a baseline and a measured period.

—

Investigation time

Awaiting an agreed definition and a source.

—

Examination preparation

Awaiting the institution's own figure.

Questions

What banks ask first

Can subsidiaries and branches be separated without running separate systems?
Yes. Org units scope who sees which book, which queues route where, and which aggregates roll up. Data isolation between tenants is enforced in the database rather than in application code, and org-unit scoping operates inside a tenant, so a group can hold its entities in one tenant with real separation, or in separate tenants where the legal structure demands it.
How does this coexist with our existing core banking system?
As a consumer of normalised transactions and customer data rather than a replacement for either. Ingest is idempotent, so a replayed batch does not double-count; invalid records are quarantined and reported rather than silently dropped; and mixed-currency input is refused rather than aggregated into a meaningless total.
What happens to our existing tuning work and thresholds?
They are re-expressed as governed configuration — proposed, validated, approved and versioned. That migration is deliberately not automatic: a threshold carried across without a documented owner and approval date is exactly the artefact that becomes indefensible two examinations later.
Can the platform file directly to the NFIU today?
The reporting engine generates report content from governed mappings and refuses to record a submission without external filing evidence. The authoritative schemas, code lists and channel procedure are supplied and accepted by your MLRO during implementation; until they are, filing stays fail-closed. We would rather state that plainly than discover the gap in UAT.

Related reading

Start the conversation

Bring your alert volume and your org chart

The useful demo for a commercial bank is not a feature tour. It is walking one of your own escalation paths — branch to regional to MLRO — and seeing what the record looks like at the end of it.

Book a demoTalk to a compliance specialist