The Scoring Loop

From Engagement Brief to Match in 90 Seconds.

Crewpath runs four independent scoring passes — domain expertise, client chemistry, utilization pressure, bench cost — then composites a ranked list your resource manager can confirm in a single click.

Start with a Retroactive Pilot
Abstract data nodes visualization representing the scoring engine data flow
Step 01 — Ingest

Engagement Brief Submitted

The process begins with an engagement brief — entered manually by a resource manager or submitted automatically via PSA webhook when a new project is opened in your system.

Client name and relationship history
Engagement type (M&A, restructuring, audit, etc.)
Duration and start date
Required skills and seniority band
Billing rate and margin target
Engagement Brief
Client
Ardent Advisory Group
Type
Operational Restructuring
Duration
10 weeks
Start
2026-07-14
Required Skills
Change MgmtFinancial ModelingStakeholder Comms
Step 02 — Score

Four Dimension Scoring Pass

Four independent scoring algorithms run simultaneously for every consultant in your roster. Each pass evaluates a different signal source. Combined, they produce a composite score that reflects both fit and firm economics.

No black box — every score is explainable down to the input signals that drove it. Partners can inspect any dimension score before confirming.

Step 03 — Rank

Ranked Output Generated

Every consultant in your roster receives a composite score. The ranked list is sorted by score and delivered to the resource manager and configured approvers.

Each row shows the overall score plus a per-dimension breakdown. Conflict flags — double-booking, ceiling breach, blacklisted pairing — appear inline.

eng_ardent_2026_07 — Ranked output • 47 consultants scored
#ConsultantDomainChemistryUtil.Score
#1 C. Worthington 917885 84.7
#2 P. Delacroix 828571 79.2
#3 A. Okonkwo 797490 74.1
#4 S. Maruyama 658055 61.4
#5 R. Kovacs 606288 58.9
Step 04 — Confirm or Override

Crewpath Proposes. Partners Decide.

Crewpath is a decision support tool, not a decision maker. The ranked list is a scored recommendation with full dimension transparency. The partner reviews it, confirms the top match, or selects a different consultant with a reason code captured in the log.

Every override is logged. Monthly calibration reports surface patterns — whether overrides cluster around a particular engagement type, client sector, or seniority band. Over time, override data feeds model recalibration for firms on the Enterprise tier, narrowing the gap between what the model proposes and what partners actually choose.

Override reason captured and logged
Override pattern visible in monthly calibration report
Enterprise: custom model retraining incorporates override data
Confirmation Interface
Recommended Match
C. Worthington
Composite score 84.7 • Senior Manager
Selecting Override will prompt for a reason code before confirming a different consultant.
Retroactive Mode

Score Your Past 12 Months of Allocations

Before committing to a scoring model, see how it would have performed on your firm's actual historical data. Retroactive Audit Mode re-runs the scoring engine against every engagement in your chosen period.

The output shows you where the model's top recommendation agreed with your committee's choice — and where it diverged. Divergence isn't always bad; sometimes it surfaces candidates your committee systematically overlooked.

Start a Retroactive Pilot
Retroactive Audit Summary
Engagements scored 142
Model agreed with committee 73%
Divergences flagged for review 38
Typical first-run results. Model calibration improves after weight tuning session.

Start with a retroactive pilot.

We run the model against your last 12 months of engagement data. See the divergence report before you commit to anything.

Request Retroactive Pilot