Outsourced Fraud Analysis: Building FCA-Grade Detection Without UK Headcount

The False Positive Problem Killing Fintech Growth

Every fraud detection system generates false positives. Rule-based systems are especially blunt — they catch patterns, but they also flag legitimate transactions that share characteristics with fraudulent ones. Industry estimates suggest that false positive rates for automated fraud rules typically fall between 60% and 70%. That means the majority of alerts your system generates are not fraud.

Each unresolved false positive has a cost. Blocked transactions mean failed payments, abandoned checkouts, and customer complaints. Left in the queue, they erode trust. Cleared too slowly, they trigger chargeback windows. And every hour a fraud analyst spends investigating a false alarm is an hour not spent identifying genuine threats.

The problem is not the detection system. The problem is human review capacity. UK fraud analysts cost £34,000+ in base salary, take 8 weeks to hire, and require platform-specific training before they can process alerts independently. In the time it takes to recruit one analyst, your alert queue has grown by thousands of unresolved cases.

How Outsourced Fraud Analysis Works in Practice

Outsourced fraud analysis is not a separate system running alongside your existing operations. In Treba’s model, dedicated analysts in Nairobi sit inside your fraud detection platform — Featurespace, Feedzai, NICE Actimize, or internal dashboards — and process alerts in real time via secure VPN.

The work covers three operational layers. Layer one is real-time transaction monitoring: flagged transactions are reviewed within minutes, genuine risks are escalated immediately, and false positives are cleared before they impact the customer. Layer two is case investigation: when genuine fraud is identified, the analyst builds a complete case file — transaction timelines, account linkage analysis, and supporting evidence — structured for your MLRO or law enforcement referral. Layer three is chargeback dispute resolution: managing the representment process, gathering evidence, preparing dispute packs, and filing within card scheme deadlines.

The shift structure is built for continuous coverage. Nairobi’s GMT+3 timezone provides near-complete overlap with UK business hours, and shift rotations can extend to 16-hour or 24/7 coverage depending on transaction volume.

Typical Coverage Model

Why Pattern Recognition Matters More Than Playbooks

The difference between a competent fraud analyst and a great one is pattern recognition. Playbooks catch known attack vectors. Pattern recognition catches the novel ones — the transactions that technically pass every rule but feel wrong when you understand the customer’s normal behaviour.

Treba’s fraud analysts are recruited from Kenya’s finance graduate pool. A BCom in Finance from the University of Nairobi covers financial systems, risk management, and statistical analysis. When combined with Certified Fraud Examiner (CFE) candidacy and daily exposure to high-volume alert queues, these analysts develop the instinct that playbooks cannot teach.

This is not an argument against automation. It is an argument for the human layer that automation requires. The best fraud detection stack combines automated rules (speed) with trained human analysts (judgement). Outsourcing the human layer to Nairobi gives you both at a cost structure that scales.

The Economics of Outsourced Fraud Analysis

A fraud analyst in London costs approximately £34,000 in base salary. Adding 13.8% employer National Insurance (£4,692), £5,000 for office space, and £4,000 for recruitment and compliance setup brings the loaded annual cost to £47,692. Through Treba’s EOR model, the equivalent analyst in Nairobi costs £12,000 per year — a saving of £35,692 per head.

The saving is not just per-head arithmetic. Because outsourced analysts cost 75% less, you can afford coverage structures that would be prohibitively expensive domestically. A 24/7 fraud monitoring function requires a minimum of 4 analysts on rotation. In London, that is £190,768 per year in loaded costs. In Nairobi, it is £48,000. The difference — over £140,000 annually — can fund better detection tooling, more sophisticated rule engineering, or a senior fraud strategy hire in London who designs the system while Nairobi executes it.

This is the real value proposition: outsourcing execution frees budget for strategy.

The Compliance Framework for Offshore Fraud Operations

Fraud data is sensitive data. Transaction details, customer information, and investigation records all fall under UK GDPR and, for FCA-regulated firms, under additional conduct requirements. Outsourcing this function to Kenya requires a complete compliance architecture — not a theoretical one.

Treba operates under the UK International Data Transfer Agreement (IDTA) for all Kenya-based data processing. Data Processing Agreements are executed per client before any data is accessed. Kenya’s Data Protection Act 2019, modelled on EU GDPR, provides a statutory baseline that aligns with UK requirements. At the infrastructure level, analysts access client systems through Virtual Desktop Infrastructure with no local data storage. The Nairobi operations centre maintains ISO 27001-aligned controls including biometric access, clean desk policy, CCTV monitoring, and network segmentation.

For FCA-regulated clients, Treba provides additional documentation: staff training records showing FCA conduct requirements completion, operational resilience assessments, and third-party risk management evidence suitable for FCA Section 166 skilled person reviews.

SAR Filing and Regulatory Boundaries

One area where the operational boundary is clear: Suspicious Activity Reports. Under the Proceeds of Crime Act 2002, SAR filing responsibility sits with the firm’s nominated Money Laundering Reporting Officer. This cannot be outsourced.

What can be outsourced is the work that precedes the filing decision. Treba’s fraud analysts prepare SAR documentation with complete evidence chains — transaction sequences, account linkage, behavioural anomalies, and risk scoring — structured for the MLRO to review and submit. The analyst does the investigative work. The MLRO makes the filing decision. This division maintains regulatory compliance while removing the investigative bottleneck.

Any provider that claims to file SARs on your behalf is either misunderstanding the regulation or misrepresenting the service. The boundary is non-negotiable.

Structuring a Fraud Analysis Pilot

• The pilot framework follows Treba’s standard 4-phase onboarding:
• Days 1–3: Discovery. Define alert categories, platform access requirements, escalation protocols, and SLA targets.
• Days 3–5: Talent Selection. Match analysts from the pre-vetted pool based on platform experience and fraud typology.
• Days 5–10: Tech & Compliance Setup. VPN configuration, platform training, DPA execution, IDTA documentation.
• Days 10–14: Nest Training. Analysts process a controlled sample of historical alerts under supervision, with QA scoring against the client’s rubric.

The recommended pilot scope is 4 weeks with 2 analysts processing a defined alert category — typically the highest-volume, lowest-complexity alert type where false positive rates are highest. This gives enough data to measure resolution time, clearance accuracy, and cost per case. If the numbers work, scale to additional categories and shift coverage.

The 30-day performance guarantee applies: if an analyst does not meet agreed standards, Treba replaces them at no cost.