How the VETR Score works
VETR scores a bid on four pillars — Value, Experience, Teaming, Responsiveness — and turns them into a win probability. Here is exactly how that number is calculated, what data it uses, and where that data goes.
Four pillars, one score
Every federal evaluation rewards the same things. VETR makes them explicit and scores each one.
Value
The measurable ROI and cost-effectiveness of your solution — priced to be credible, not just cheap.
Experience
Relevant, recent past performance that proves you have done this work and delivered it well.
Teaming
The partnerships and certifications that fill capability gaps and de-risk delivery for the government.
Responsiveness
Full, on-time compliance with every stated requirement — one missed mandatory is the fastest way to get eliminated.
Honest on day one. Smarter as you grow.
VETR never pretends to know more than it does. The method it uses depends on how much of your own history it can learn from.
Transparent heuristic
Before your account has enough history to learn from, your win probability is simply the average of your four VETR pillar scores. No hidden weighting, no invented factors — a plain, honest baseline you can reason about.
A model trained on your own wins and losses
Once your organization has logged 20+ won/lost outcomes, VETR trains a calibrated model on your own history. It reports the specific factors moving each estimate up or down — so you see what is driving the number, not just the number.
When the model is in use, it shows the top factors driving your estimate — each with its own direction and magnitude — so you know exactly which lever to pull next.
What the model learns from
The model is trained on structured signals only. Your proposal narrative, past-performance write-ups, and any controlled unclassified information (CUI) never enter a feature vector.
- The four VETR pillar scores and their composite
- Compliance completeness — requirements met vs. outstanding
- Teaming depth — partners, subcontractors, discriminators
- Experience depth — count of relevant past performance
- Price positioning — bid-to-estimate ratio and margin
- Public solicitation context — agency, NAICS, set-aside type
Where your data goes
Nowhere it shouldn't. The scoring math runs in-boundary — pure computation, no external service. Any AI-written explanation is generated inside the same AWS GovCloud authorization boundary.
No external-LLM egress
Your proposal content is never sent to a commercial AI provider.
Per-organization, isolated
Your model learns from your history alone — never another tenant's.
See your starting point
The free 5-minute readiness check scores you on all four VETR pillars — no login, no card.
Value · Experience · Teaming · Responsiveness