Skip to content

Win Probability

GovConHouse shows several numbers for a notice or a pursuit. They answer different questions and use different inputs, so one notice can carry three different numbers. None of them is a forecast that you will win. Use them to decide where to spend your time.

  • Fit score (the badge on the Opportunities list): how closely the notice matches your company profile, 0 to 100. NAICS 35 points, set-aside 25, agency history 20, place of performance 10, your search words 10. A criterion the platform cannot judge is left out. A set-aside you are not certified for caps the score at 35. No AI.
  • Bid / No-Bid Cockpit (the full notice page): eight checks with weights that add to 100: eligibility, scope fit, past performance, competition and incumbent, clearance, size against capacity, teaming need and time to respond. Each check is Green, Amber or Red with its reason and source. The total ends in Pursue (70 and above), Get more information (50 to 69) or Pass (below 50). A red gate, such as a set-aside you do not qualify for or a closed notice, forces Pass. The “Closest to flipping” line names the one change that would move the result. No AI.
  • Feasibility estimate (the preview next to a notice, labeled Estimate): five factors, each 0 to 25: win probability, cost competitiveness, compliance readiness, capacity and strategic alignment. Go at 70 and above, Review from 40 to 69, No go below 40. It starts every factor at half credit or a little above and moves on a few facts from your profile, so a thin profile lands in the middle. It has no set-aside gate, so it can read more favorable than the Fit score or the Cockpit. No AI.
  • P(win) on the pipeline: the probability you enter. If you enter none, an open pursuit gets a stage default (Identified 10 percent, Qualifying 20, Bidding 35, Submitted 40), labeled “stage default” or Estimate. The weighted value is the pursuit value times P(win).
  • PWin ratings: your team’s own 0 to 10 ratings on seven categories. See PWin Scoring.

This page uses a model trained only on your own decided bids, one model per company. With fewer than 30 decided bids no model is trained, and each bid shows “Insufficient history” and is left out of the summary numbers. It is a model, not a language model, and not a promise of an award.

They do not know who else is bidding, your price or the government’s budget. The Cockpit’s competition check reports how many offers similar awards drew, which is a count, not a probability. They read only your company profile, your records and public award data.

Try this in the platform