How matching works

Two stages. First, hard filters narrow the candidate pool to providers who can structurally serve the RFQ. Then, soft scoring across twelve dimensions ranks the providers who actually respond. Admin curation sits between the two stages.

01Buyer submits RFQ
02Hard filters
03Admin curation
04Provider responses
05Soft scoring
06Comparison output
01
Stage one

Hard filters

A provider either passes a hard filter or they are excluded from the candidate pool for the RFQ. These are deterministic — they read directly from the provider's capability profile and the buyer's RFQ. No judgment, no scoring.

01Currency support

Provider must support both the source and destination currency of every corridor on the RFQ.

02Corridor coverage

Provider must reach every destination country either directly or via a declared correspondent.

03Payment type

Provider must support at least one of the buyer's preferred payment rails for each corridor.

04Monthly volume floor

Buyer's monthly volume must clear the provider's stated minimum (per provider profile).

05Segment exclusions

Provider's declared exclusions (e.g. adult, firearms, dual-use) must not overlap with the buyer's customer base.

02
Between stages

Admin curation

Railflows admin reviews the candidate pool that survives hard filters and decides which providers actually receive the anonymised RFQ. This is the platform's editorial layer — providers obviously fit on paper don't always fit in practice, and providers borderline on the filters sometimes warrant inclusion based on a recent corridor expansion or capability change not yet reflected in their profile.

Admin curation is logged in the audit trail. Buyers can see which providers were considered and which made the cut. Providers can see when they were considered for a match and what happened.

03
Stage two

Soft scoring (12 dimensions)

Once a provider responds, their response is scored across twelve dimensions. Each dimension is a 0–100 score with a weight; the weighted average becomes the provider's overall fit. Weights below reflect a typical RFQ — they shift based on the use case (treasury FX leans pricing-heavy, payroll leans coverage-heavy).

Coverage fit
22%
Pricing (composite)
12%
Risk fit
10%
Settlement speed
10%
FX spread
8%
Onboarding feasibility
8%
Transaction fees
5%
Prefunding
5%
Credit availability
5%
API maturity
5%
Support / SLA
5%
Response engagement
5%
04
Boundaries

What scoring does and does not do

Fit scoring does
  • Evaluate provider response quality against the specific RFQ.
  • Weight coverage, pricing, settlement, risk, onboarding and engagement.
  • Use the buyer's own corridors and volume to compute coverage match.
  • Penalise non-responses and incomplete responses.
Fit scoring does not
  • Rank by payment. No provider can buy ranking placement. Ever.
  • Guarantee buyer selection — final selection is the buyer's decision.
  • Replace provider due diligence — buyers run their own checks before signing.
  • Imply regulatory endorsement of any provider's licences.
05
Correction loop

Providers can challenge a score

If a provider believes their score on a specific RFQ misrepresents their capability — because of recent profile changes, a corridor expansion, or because a dimension was scored on stale data — they can raise a correction request. Railflows admin reviews, and if warranted, re-scores. Re-scoring is logged in the audit trail and the buyer is notified if their comparison changes.

See it in practice

Sample RFQ → sample comparison

The two-stage flow is easiest to understand when you see the input and the output side by side.