Vendor Scorecard for AML and Fraud Platforms
A weighted scoring template for compliance teams running a vendor selection. Free spreadsheet, no fields to fill in.
Most vendor comparisons collapse under their own scorecard. Either every criterion is weighted equally, which selects a platform that is adequate at everything and excellent at nothing, or the weightings get set after the demos, by which point they have quietly reshaped themselves around whatever the strongest presentation emphasised.
This template fixes the order of operations. You set weightings first, score against a fixed scale second, and reconcile total cost third.
What is in the workbook
Weightings. Eight criteria: detection approach, transaction coverage, decision speed, configurability, alert workflow, investigations, explainability, and operational fit. Adjust each weight to your institution, with a validation check that the total reaches 100%. Suggested starting weights are included for six institution types, from a scaling fintech to a credit union to a cross-border payment provider.
Scorecard. Forty-one sub-criteria across the eight categories, scored 0 to 3 for up to four vendors. Category averages, weighted scores, and the final total calculate automatically. Eight items are marked DEMO: these cannot be scored above 2 on a written response alone, because they are the things a vendor can describe convincingly and still be unable to show.
Total Cost. Six cost lines, of which vendors quote one. The template names where each of the other five comes from, keyed to specific scorecard items. Line 4, analyst headcount implied by projected alert volume, is usually the largest and appears in no proposal. A worked example shows how two vendors quoting identical licence fees can differ by six figures a year.
Evidence Reference. Published vendor documentation is mapped item by item to the scorecard to demonstrate the level of evidence a compliance team should expect during evaluation. The workbook uses Flagright as one reference point, reviewing publicly documented capabilities such as transaction monitoring, configurable rule creation, rule simulation, investigations, screening, and regulatory workflows against the relevant criteria. It also identifies questions that public documentation alone cannot resolve and therefore need to be tested during a live evaluation. Fourteen other vendors commonly evaluated in this category are described briefly and factually. Nothing in the workbook is a ranking.
The scoring scale
| Score | Meaning |
| 0 | Not supported. |
| 1 | Supported, but requires vendor involvement, engineering work, or a paid change request. |
| 2 | Supported and self-service, claimed by the vendor but not demonstrated. |
| 3 | Supported, self-service, and demonstrated live during evaluation. |
The gap between 1 and 2 is where most of the real difference between platforms sits, and it is the gap a feature matrix cannot show you. A vendor answering “yes, we support that” may mean either.
The eight items to demand live
Written responses cover the rest. These eight need demonstration:
- p99 latency and sustained throughput under load, not average figures.
- A non-technical person building and deploying a rule, timed.
- Alert volume and implied analyst headcount projected at your transaction count, with the working shown.
- A rule backtested against historical data showing projected alert volume before deployment.
- One case walked end to end: alert, evidence, escalation, approval, decision, filing, submission receipt.
- A decision from two years ago reconstructed with the rule logic in force at the time.
- Go-live in days named to a specific customer at your volume, with engineering hours required from you in writing.
- What changes commercially and operationally when you double volume or add a jurisdiction, in writing.
A vendor who produces all eight quickly is telling you something about the product. A vendor who produces two is also telling you something.
Two notes on using it well
Compare within an archetype. Shortlist three or four vendors from at most two adjacent categories. An enterprise financial crime suite and a lean cloud platform are optimised for different constraints, and putting both in one scorecard produces a total nobody can reconcile.
Read the category rows, not just the total. A weighted total is a conversation starter. The useful information is where each vendor is strong and whether any category scoring below 2.0 is one you weighted heavily. A platform that wins on total while scoring 1.2 on explainability is not the platform you want in an examination.
