[ Case Study ]
Digi Assist: account opening for the Nigerians the branch never reached
01[ Context ]
The Central Bank of Nigeria's tiered KYC framework was designed to expand financial access. Tier 3 accounts unlock daily transaction limits of up to N5 million and unlimited balances, but opening one still required in-person visits, physical documents, and manual verification workflows that took hours and sometimes days. For the 26% of Nigerian adults locked out of formal banking, the regulation existed. The access did not.
As Lead Technical Business Analyst, I was responsible for translating the vision for a fully digital, straight-through Tier 3 account opening experience into a working system: four account types, six stakeholder roles, three external integrations, and a regulatory environment that left no room for shortcuts.
02[ Business Problem ]
Traditional Tier 3 account opening at Nigerian banks meant branch visits, extended queues, physical document submissions, and manual compliance checks that stretched the process across hours or days. The CBN's Enhanced Verification requirement for Tier 3 made the process even more demanding: ID validation, biometric checks, and document review before an account could be activated.
This created a real gap. The regulation was there. The demand was there. But the infrastructure to deliver Tier 3 accounts digitally, securely, and in a way that actually worked for customers outside major urban centres did not exist.
03[ Constraints ]
- CBN tiered KYC frameworkTier 3 accounts carry the highest transaction privileges, which means the verification bar is the highest. Every step in the onboarding flow had to be sequenced correctly and logged for audit, from BVN/NIN validation and OTP confirmation to document upload, biometric check, and sanctions screening.
- NIBSS as a hard dependencyReal-time validation against the national identity database was non-negotiable. Any failure mode in the NIBSS API layer had to degrade gracefully without allowing an unverified account to be activated.
- AML/CFT complianceSanctions screening could not be a downstream check. The PND mechanism had to guarantee that no account could be used until compliance review was complete, protecting the bank’s regulatory position on every single account opened.
- PND management complexityThe logic for placing, escalating, and lifting PNDs was different across account types and exception scenarios. Getting this right required precise specification and close collaboration with the Compliance and KYC teams.
- Data privacyData privacy obligations were formally acknowledged and built into the integration design across all external touchpoints, covering how customer identity data was handled, stored, and transmitted between the platform and national identity systems.
04[ Stakeholder Landscape ]
Every onboarding flow crossed six roles: Staff, Bot, Compliance Officer, KYC Officer, Onboarding Officer, and the Automated Verification and References (AVR) system. Around that operational core sat Compliance, KYC, Legal, solution architects, engineering teams, data scientists, and external technology vendors for biometrics, OCR, and identity validation.
05[ Research ]
The work started with a deep regulatory analysis of the CBN's tiered KYC framework, run alongside as-is process mapping with the stakeholders who lived the manual flow every day. I mapped every Enhanced Verification requirement for Tier 3 against what was technically feasible, identified the compliance gaps in the existing onboarding flow, and translated that into requirements the development team could act on.
A feasibility study grounded the build decisions. Getting this analysis right at the start was critical; every design decision downstream depended on it.
06[ Strategy ]
The strategy was straight-through processing with compliance built into the core rather than bolted on after. Automated sanctions screening was embedded in the BOT layer, so every account was screened before any human reviewer touched it. Accounts that passed were created immediately, with documents pushed to the core banking, document management, and workflow systems simultaneously. Accounts that failed were flagged with a PND and routed to the Compliance Officer.
That inversion, automation first with human review reserved for exceptions, is what collapsed account opening from hours to minutes without lowering the verification bar.
07[ Options Considered ]
- option 01Digitise the form, keep manual verificationThe incremental path: an online front end feeding the same human-driven KYC queue. It would have improved the application experience while leaving the real bottleneck, manual review and physical document handling, fully intact.
- option 02Straight-through processing with human review on every accountSafer on paper, but a compliance officer in the loop for every single account caps throughput at the size of the review team and reintroduces the hours-long turnaround the project existed to remove.
- option 03Straight-through processing with automated screening and PND gatingThe chosen path. Automated sanctions screening in the BOT layer clears the clean majority instantly, while the PND mechanism guarantees flagged accounts cannot transact until a human completes the review. Speed for customers, control for the bank.
08[ Trade-offs ]
- Automation depth over specification simplicityScreening every account in the BOT layer meant specifying every branch of the logic, including the False Violation path, across all four account types. The specification burden was heavy, but it is what made minutes-long account opening compatible with the CBN’s verification bar.
- Real-time identity validation over offline batch checksBuilding the KYC process on live NIBSS BVN and NIN validation made the platform dependent on an external API in the critical path. The cost was designing fallback behaviour for every edge-case response; the payoff was that no account could be activated on stale or unverified identity data.
- Four tailored flows over one generic flowTier 1, Tier 3, Current, and Salary accounts each got their own swim lane design with distinct document requirements, PND logic, verification triggers, and exception paths. More design work up front, but it kept the lean Tier 1 journey lean instead of dragging every customer through the heaviest flow.
09[ Delivery Process ]
- 01. Regulatory analysis and gap mappingMapped every CBN Enhanced Verification requirement for Tier 3 against technical feasibility, identified compliance gaps in the existing onboarding flow, and turned the findings into actionable requirements.
- 02. Swim lane process designDesigned four distinct swim lane flows covering Tier 1 Savings, Tier 3 Savings, Current Account, and Salary Account onboarding, each mapped across the six operational roles with its own exception handling paths.
- 03. NIBSS integration workstreamDefined API contracts, data validation rules, and fallback behaviour for real-time BVN and NIN validation with solution architects and engineering, covering both customer-facing validation and compliance-side requirements.
- 04. Sanctions screening and PND logicSpecified the automated screening flow in the BOT layer, the simultaneous document push to the core banking, document management, and workflow systems on pass, and the PND flag with Compliance Officer notification on fail, across every branch including False Violations.
- 05. Biometrics, OTP, and OCR integrationOversaw the liveness detection and facial recognition layer, OTP verification via BVN-linked phone and email, and OCR document capture, each with its own vendor coordination and failure mode behaviour specified in the SRS.
- 06. KPI framework and post-launch iterationDefined post-launch measurement covering activation rates, verification failure rates by step, processing times, compliance review turnaround, and AVR completion rates, then used the reports to drive iterative optimisation in the months after launch.
10[ Technical Architecture ]
The platform hangs off a real-time identity backbone. NIBSS BVN and NIN validation anchors every account type, with OTP confirmation over the BVN-linked phone and email, OCR capture from government-issued IDs, and biometric liveness and facial recognition layered on top. The BOT layer runs automated sanctions screening on submission; clean accounts are created with documents pushed simultaneously to the core banking, document management, and workflow systems, while flagged accounts carry a PND until compliance review completes. AI/ML models watch for transaction anomalies and feed AML monitoring tied to the CBN's AML/CFT mandate.
Core Banking
Existing core banking platform (integrated, not replaced)
Document Management
Document management system for KYC artefacts
Workflow Management
Workflow orchestration engine
National Identity
NIBSS (BVN and NIN real-time validation)
Biometrics
Liveness detection, facial recognition
Document Processing
OCR (automated data capture from government IDs)
Fraud Detection
AI/ML models (anomaly detection, AML monitoring)
Communication
Automated OTP via BVN-linked SMS and email
Project Tracking
Jira + Confluence
11[ Outcomes ]
The platform launched as Nigeria's first truly end-to-end on-the-go account opening solution. Over 20,000 accounts were opened across Tier 1 and Tier 3 within the first three months. Account opening time dropped from hours to minutes. Operational costs fell by approximately 15% through automated KYC and reduced manual intervention. The platform delivered financial access to a segment of the Nigerian population that the existing infrastructure had consistently failed to serve.
The project was recognised internally at Stanbic IBTC with the Beyond Excellence Award for Innovation and Teamwork.
Artefacts delivered
12[ Metrics ]
13[ Lessons Learned ]
- Regulatory analysis is the foundation, not a checkpoint. Mapping the CBN framework before any design work meant every downstream decision inherited a correct compliance posture instead of retrofitting one.
- Automation and control are not opposites. The PND mechanism is what made automated screening acceptable to Compliance: the system could move fast precisely because flagged accounts were provably unable to transact.
- Failure modes deserve the same specification rigour as happy paths. The edge-case behaviour of NIBSS responses, biometric checks, and OCR capture is where a straight-through platform either holds or leaks, and specifying it precisely up front is what kept unverified accounts out.