Why KYC / AML Vendor Selection Is Structurally Difficult
Every KYC / AML vendor claims high accuracy, ECB / EBA compliance, seamless integration, and competitive pricing. The claims are impossible to verify from brochures and sales calls alone — you need live data, and the only way to get live data is to run a pilot, which takes months and costs money. And by the time you have pilot data from three vendors, your original selection criteria may have changed because the ECB / EBA issued a new guideline, a new vendor entered the market, or your volume grew past the tier that attracted the pricing you were quoted.
The Onboarding Head AI resolves this by maintaining a continuously updated evaluation framework that draws on industry benchmark data, public regulatory action records, peer lender intelligence (where available through industry bodies), vendor-published SLA performance data, and the institution's own pilot results where they exist. It does not rely on vendor-provided claims — it cross-references every claim against independent evidence before it enters the evaluation.
The Eight Evaluation Dimensions
V-KYC / AML Pass Rate
What percentage of borrowers who initiate a V-KYC / AML session complete it successfully? Benchmarked against segment type — salaried urban differs from self-employed rural.
Weight: 20% of total scoreECB / EBA Compliance Record
Has the vendor received any ECB / EBA advisory or been cited in inspection observations? Are their processes current against the latest KYC / AML ECB guide / EBA guideline amendments?
Weight: 18% of total scoreDocument OCR Accuracy
Accuracy rate on national ID / eIDAS, PAN, and driving licence extraction across varied image quality levels. Tested against a standardised document quality benchmark set.
Weight: 15% of total scoreLiveness & Spoof Detection
False accept rate on photo spoofing and screen-replay attacks. Tested across device types. Minimum acceptable standard: FAR below 0.1% for financial-grade KYC / AML.
Weight: 15% of total scoreAPI Uptime & SLA
Contractually committed API uptime, observed uptime from public status pages and peer data, average response latency at peak load, and incident recovery time record.
Weight: 12% of total scoreData Localisation & Security
Data storage within the EU as required. ISO 27001 and SOC 2 certifications. Penetration test recency. GDPR compliance posture and processor agreement terms.
Weight: 10% of total scoreIntegration Complexity & Speed
SDK quality, API documentation completeness, median time-to-integration from peer lender data, quality of sandbox environment, and post-go-live support SLA.
Weight: 6% of total scoreTotal Cost of Ownership
Per-verification pricing across volume tiers, cost of failed verifications, overage pricing model, renewal terms, and total cost modelled at the institution's 3-year volume projection.
Weight: 4% of total scoreThe Vendor Scorecard: What the AI Produces
The output of the Onboarding Head AI's evaluation is a structured vendor scorecard — not a summary of what vendors told the AI, but a scored matrix built on evidence from each dimension with sources cited and confidence levels attached. The scorecard is the input to the human decision, not a replacement for it. But it is a complete, defensible input — one that the institution can present to its board, its regulators, and its legal team as the basis for a vendor selection decision.
| Vendor | V-KYC / AML Pass Rate | ECB / EBA Compliance | OCR Accuracy | Liveness / Anti-Spoof | API Uptime | Data Security | Total Score | Verdict |
|---|---|---|---|---|---|---|---|---|
| Vendor A — DigiVerify | 91.2% | Clean | 98.4% | FAR 0.04% | 99.94% | ISO+SOC2 | 87/100 | Recommended |
| Vendor B — IDFlow | 86.7% | 1 advisory | 97.8% | FAR 0.09% | 99.91% | ISO+SOC2 | 79/100 | Conditional |
| Vendor C — KlearKYC | 83.1% | Clean | 95.2% | FAR 0.06% | 99.78% | ISO only | 74/100 | Alternative |
| Vendor D — VeriCheck | 78.3% | 2 advisories | 94.6% | FAR 0.18% | 99.61% | ISO only | 61/100 | Not Recommended |
The Recommendation Document: What the AI Delivers to the Decision-Maker
The Onboarding Head AI's vendor recommendation is not the scorecard alone. It is a structured recommendation document that includes the scorecard, a narrative recommendation with rationale, a risk assessment for the recommended vendor (what could go wrong and what the mitigation is), a contract negotiation brief (which terms to push on, which SLA clauses to insist on, what the benchmark pricing should be at the institution's volume), and a transition plan if the recommendation involves switching from an existing vendor.
For Vendor A in the scorecard above, the recommendation document would note the strong V-KYC / AML pass rate and clean regulatory record as the primary selection drivers, flag the slightly higher per-verification cost compared to Vendor C as an acceptable premium given the 8-percentage-point pass rate advantage (worth approximately €4.2 hundred thousand per month in recovered disbursements at the institution's volume), and recommend negotiating a performance SLA clause that allows fee reduction if the pass rate falls below 88% in any rolling 30-day period.
This is not a generic vendor comparison. It is a commercially grounded recommendation specific to the institution's volume, segment mix, and regulatory context — produced in the time it would take a product manager to schedule the first vendor briefing call.
The KYC / AML Vendor Is the First Experience a Borrower Has of Your Brand
A V-KYC / AML session that fails — due to a confusing interface, an unreliable connection handler, or a spoof detection system that false-rejects genuine borrowers — is not a technology incident. It is a brand incident. The borrower who was rejected by a poor KYC / AML vendor does not blame the vendor. They blame the lender. The Onboarding Head AI ensures the vendor you choose reflects the experience you intend to deliver.
