Are these case studies from named customers?

No. Per our compliance policy, all case studies are anonymized by category (neobank, digital lender, embedded finance platform). Customer names are protected under NDA; aggregate ROI figures are independently audited and published in our annual State of Fintech Risk report.

How were the ROI figures in these case studies calculated?

Each pilot ran a 90-day A/B framework against the customer's incumbent system. Baselines were captured over the prior 90 days; deltas were measured on identical transaction cohorts. Full methodology notes and supporting evidence are available under NDA within 24 hours of a demo request.

Can I get the full case-study pack including raw data?

Yes. Booked demos receive a confidential case-study pack containing raw model outputs, deployment timelines, and the methodology note. Audit-grade documentation (SOC 2 Type II report, ISO 27001 certificate, and PCI-DSS Level 1 attestation) is also available under NDA within 24 hours.

What is the typical time from demo to first results in our environment?

Median model deployment time is 11 days for a single use case (fraud, credit, or KYC), compared to the 14-week industry benchmark reported by Gartner in 2024. Multi-use-case rollouts typically complete within 6–8 weeks.

Do you publish aggregate customer ROI across all deployments?

Yes. Across 380+ fintech teams in production, customer-reported average ROI is 5.8x within the first 12 months of deployment, per the 2024 Customer Satisfaction Benchmark. We publish anonymized aggregates in our annual State of Fintech Risk report, now in its fourth edition.

pulsaf5.com · all systems nominal en-US · v4.2 dossier

Case Studies & ROI · 2024 deployment archive

Three quantified deployments. One ROI dossier a CFO can take to a budget meeting.

Pulsaf5 has been deployed against fraud, credit, and KYC workloads at 380+ fintechs since 2019. The three studies below document, with full numbers and methodology, what the platform does to the metrics that matter: false-positive rates, approval rates, onboarding throughput, and time-to-decision. Every figure is reproducible from the raw event logs under NDA.

Book a 20-minute demo Read the methodology

RISK-SCORE TIMELINE · 24H WINDOW SCHEMATIC

00:0006:0012:0018:0023:59

p_fraud

DECISIONS

REVIEW Q

Score spike Decision event Manual review

Neobank · Fraud · 14-week pilot

A top-10 neobank cut fraud false positives by 23% while expanding detection coverage.

A tier-1 digital bank with 6.4M monthly active accounts deployed Pulsaf5's adaptive scoring alongside their incumbent rules engine. The pilot covered 100% of card-not-present transactions for 14 weeks, with the legacy engine scoring in parallel as a holdout group.

False-positive reduction −23% vs. baseline rules engine

Fraud-loss prevented $8.4M annualized, net of false-positive cost

Detection coverage +6.1pp recall on confirmed-fraud cohort

Time to deploy 11 days from API key to first scored event

What changed

Scoring layer — Pulsaf5 sat between the issuer processor and the rules engine, returning a calibrated score on every CNP attempt in under 35 ms p99.
Review queue — Reviewer load dropped from 4.2% to 3.1% of total transactions while the genuine-fraud capture rate rose 6.1 percentage points.
Operating model — The bank's risk-ops team rebuilt policy on top of Pulsaf5's SHAP attributions; no black-box overrides were introduced in production.

How it was measured

Holdout — A 9.8% random traffic slice continued on the incumbent rules engine for the full 14 weeks.
Backtest — Pulsaf5's scores were re-run against the prior 12 months of chargebacks to confirm in-sample lift held out-of-sample.
Audit trail — Every score, feature snapshot, and final decision is available in the raw event logs for 24 months.

Digital lender · Credit decisioning · 90-day deployment

A digital-first lender lifted thin-file approvals 17% while keeping 30-day delinquency flat.

A consumer lender originating personal loans through an embedded marketplace deployed Pulsaf5's credit-scoring surface to expand into the thin-file segment the bureau-only model had been declining since 2022. The 90-day deployment window covered 412,000 unique applications.

Thin-file approvals +17% absolute lift on previously-declined segment

30-day delinquency +0.2pp inside the lender's 0.4pp tolerance band

Originated volume $214M incremental in the 90-day window

Manual underwriter hours −38% cases routed back to manual review

What changed

Decisioning surface — Pulsaf5 returned a 0–999 score plus five reason codes per applicant, consumed by the lender's existing LOS.
Segment unlocked — The thin-file cohort (no bureau hit, < 3 trade lines) moved from a hard decline to a scored-decide lane with calibrated policy limits.
Loss-rate guardrails — Hard exposure caps and a daily re-cut on the score distribution kept loss within the 0.4pp tolerance band for every weekday of the pilot.

How it was measured

Champion–challenger — Pulsaf5 ran in shadow mode for the first 21 days; decisions were compared against the legacy score before any policy was changed.
Cohort tracking — Approved applications were bucketed by score decile and tracked to 30, 60, and 90 days of performance.
Re-validation — A weekly backtest refreshed the PSI and CSI on the active population; no breach of the lender's drift thresholds in the pilot window.

Embedded finance · KYC orchestration · Platform deployment

An embedded-finance platform raised onboarding pass-through 12% and shortened step-up time by 41%.

A B2B platform onboarding SMB merchants on behalf of three partner banks consolidated four discrete KYC vendors behind Pulsaf5's orchestration layer. The deployment covered 1.8M merchant applications over 6 months across the platform's full partner network.

Auto-pass rate +12pp applications cleared without manual touch

Step-up time −41% median end-to-end on flagged cohort

Vendor count 4 → 1 single Pulsaf5 contract covering all partners

Audit-pack assembly 6× faster SOC 2 evidence compiled in hours, not days

What changed

Orchestration layer — Pulsaf5 selected the cheapest sufficient vendor per application, cutting KYC cost-per-onboard by 19% on the flagged cohort.
Step-up routing — Applications that previously fell to a 72-hour manual queue now resolve in under 42 minutes median via the integrated document + liveness loop.
Partner governance — Each partner bank's policy is enforced as a separate overlay on the shared orchestration API; no cross-leakage between cohorts.

How it was measured

Per-partner counterfactuals — The platform reproduced each partner's pre-Pulsaf5 vendor stack in shadow mode for 30 days to anchor the lift.
Funnel instrumentation — Drop-off was tagged at every vendor handoff to isolate Pulsaf5-attributable gains from network-attributable ones.
Regulator-facing pack — Every decision ships with a reproducible evidence trail accepted by the platform's partner-bank compliance teams.

Aggregated ROI · across all three deployments

The numbers a head of risk or CFO can model against their own book.

Customer-reported outcomes from the deployments above, normalized to a 12-month operating window. The right column states the underlying assumption so a reviewer can swap their own figures in.

5.8× Average first-12-month ROI Self-reported across the active customer base.

9× Faster time-to-decision Versus the customer's pre-Pulsaf5 baseline.

23% False-positive reduction Median across fraud pilots at three top-10 neobanks.

11 days Median model deployment Versus the 14-week industry benchmark (Gartner, 2024).

Figures aggregate the three case studies on this page plus the wider Pulsaf5 customer base of 380+ fintechs. ROI is self-reported by the customer; Pulsaf5 does not guarantee parity in any individual deployment.

Methodology & evidentiary standard

How these results were measured — and what's available under NDA.

Every figure on this page is drawn from a documented Pulsaf5 deployment. The default evidentiary standard is a 30-day shadow-mode baseline followed by a champion–challenger production rollout with cohort-level tracking to the customer's preferred outcome horizon (30, 60, and 90 days for credit; full chargeback windows for fraud). Population Stability Index and Characteristic Stability Index are recomputed weekly on the active score distribution; breach thresholds are defined per-customer and enforced in the joint runbook.

Pulsaf5 is SOC 2 Type II, ISO 27001, and PCI-DSS Level 1 certified. Full audit reports are available under NDA within 24 hours of a verified request from a prospect's security or compliance lead. A 41-page methodology PDF covering feature construction, calibration, and fairness testing on the 4.2-billion-transaction training corpus is available to qualified teams on the same channel.

Two constraints worth stating up front. First, Pulsaf5 does not publish named-customer outcomes; the three deployments above are anonymized by category so that aggregate numbers can be shared without naming the deploying institution. Second, no ROI figure on this page is a guarantee — every deployment is governed by the customer's own data, policy, and risk appetite. The 5.8× median ROI is a self-reported central tendency, not a contracted outcome.

Book a 20-minute demo Read the trust & research charter