The evidence
How we prove it
Measured isn’t validated. Explainable isn’t auditable. This is the methodology behind those two claims: the numbers, the loop that closes them, and what a real audit record looks like.
By the numbers
What we can already show, and what’s still being finalized
Some of this is settled fact. The validation figures are still in review before we put an exact number behind them, so we’re labeling those honestly rather than rounding to something that sounds finished.
of hiring outcome data
Drawn from millions of candidates who proved their skills through real, job-related work, not a resume.
purpose-built models in production
Trained specifically for skills assessment, then refined against real hiring outcomes ever since.
named enterprise customers
Lumen, OneMain Financial, Kroll, Rentokil, the NHL and Australia Post make hiring decisions on this data today.
score-to-outcome correlation
The headline validation number: how well an assessment score predicts on-the-job performance, by role family.
candidates tracked to outcome
How many hires we followed for six to twelve months to check the prediction against what actually happened.
segments validated
The role families and industries where we can currently show the loop closed end to end.
“In validation” means the underlying data exists and is being checked before we publish a figure. We’d rather show you the honest state of the work than a number we can’t stand behind later.
The loop, in full
How each step actually works
The manifesto compresses this into four lines. Here’s what sits behind each one.
Prove the skill
Candidates complete realistic, job-related work sampled from what the role actually requires, not a personality quiz or a resume screen.
MethodAssessment content is built with subject-matter experts per role family and reviewed for adverse impact before it ships.
Validate the score
We track who scored well against how they actually performed on the job, six to twelve months later.
MethodOutcome data comes from customer-reported performance signals, tied back to the original assessment score for that hire.
Explain the call
Every score ties back to the rubric that produced it and, where a model is involved, which model made the call.
MethodHiring managers and candidates can see the specific criteria behind a decision, not just a pass or fail label.
Audit the record
The decision is recorded in a form that still makes sense to a candidate, legal, or an auditor months later.
MethodA sample of what that record looks like is below.
What a real record looks like
Explainable in the moment. Auditable months later.
A defensible hiring decision isn’t a gut call with a score attached. It’s a record that still holds up when a candidate asks why, when legal asks for the file, or when an auditor asks for the method.
The panel shows the shape of that record. It’s a composite example built to illustrate the fields, not a real candidate’s data.
Audit record
Illustrative sample- Assessment
- Senior Support Engineer — troubleshooting simulation
- Skill measured
- Root-cause diagnosis under a live ticket queue
- Rubric criteria
- 4 weighted criteria, scored independently
- Score
- 86 / 100 — Strong hire
- Model used
- Vervoe assessment model v-support-3, logged by version
- Reviewed by
- Hiring manager sign-off, timestamped
- Retention
- Full record retained for the customer’s compliance window
Ask us for it
This is what we mean when we say trust has to be provable.
If a vendor can’t show you the rubric, the model, and the outcome data behind a score, they’re asking you to trust a hunch dressed up as technology.