How Impact Scoring Works
A look inside the system that turns dense policy text into personalized impact scores.
What we analyze
Every bill and executive order that enters Govbase goes through a multi-stage AI pipeline. The system reads the full text, identifies who is affected, and scores each impact across four dimensions. The result is a structured breakdown that captures not just whether a policy affects a group, but how much, how likely, and for how long.
Impact scores are generated by AI and should be treated as informed estimates, not certainties. We continuously refine our models and welcome feedback.
Four dimensions
Each impact is scored from 1 to 5 on four independent dimensions.
Benefit or harm
Beyond the four dimensions, each impact also receives a sentiment rating on a scale from −5 (devastating harm) to +5 (transformative benefit). This captures whether a policy helps or hurts the affected group.
The personal score
Govbase combines the four dimensions and sentiment into a single number for each impact. The favorability score shown in the app runs from 0 to 100 and represents how positively or negatively recent policy has affected a user's followed groups.
Users set up a profile by following tags that describe them, like veteran, student, or small business owner. Govbase then aggregates the sentiment-weighted scores across all matching policies to produce a personal favorability score. A score of 50 means the net impact is neutral. Above 50 means recent policy has been more beneficial; below 50 means more harmful.
Who we track impacts for
AI generates impact analyses for these groups on every policy. Follow the ones that apply and Govbase filters everything automatically.
In addition to these tags, every policy is analyzed for impact on all 50 states, D.C., and Puerto Rico.