The yardstick for AI in government - calibrated for your agency type and your systems of record.

How we test, score, and publish

Yardstick Research is an independent software research and consulting agency for B2B AI tools. We test the tools ourselves, score them on outcomes that matter, and publish the results. Methodology in plain sight, so any council, board, or oversight committee can check our work. For state and local agencies, county and municipal governments, federal agencies, public-safety and justice agencies, and special districts, we weight Data Readiness and Team & Workflow heavily because in government the integration spine (the platform of record, cross-department data integration, GIS, and the authorized cloud boundary) decides what AI you can actually deploy, and the workforce that owns the workflow - much of it civil-service and union-represented - decides whether it sticks. Here's how that actually happens:

1. 01

   We evaluate every government vendor on this list using public information and free-tier hands-on.

   Our researchers evaluate each vendor on the list using a defensible mix of inputs: vendor documentation and pricing pages, free-tier or trial-seat hands-on where the vendor offers one, video walkthroughs, third-party reviews (G2, Capterra, Gartner Peer Insights), published agency case studies, practitioner discussion (GovTech, Route Fifty, StateScoop, GovLoop), authorization listings (the FedRAMP Marketplace, StateRAMP / GovRAMP), and recent funding and news coverage. Where we can sign up and exercise the product directly, we do, and grade the output against a sample workflow: in the government case, a permit-intake-and-completeness pass through a tool like Accela or GovWell, a 311 resident-request cycle through CivicPlus or Granicus, a records-search query through Peregrine, or a budget or procurement-document task through OpenGov or Unison Global. We do not pay for paid tiers and we do not run a held-out benchmark through every tool. Both are cost-prohibitive at the scale this guide covers.

   Every claim in a tear-sheet is labelled MEASURED (free-tier hands-on observation, or output graded against a sample workflow), ESTIMATED (cost-per-seat efficiency derived from the vendor's pricing page and feature limits), or CITED (vendor-published or third-party benchmark, with the source linked).

2. 02

   We score on outcomes buyers care about, with weights we publish.

   Vendor decks sell features. Public-sector teams actually buy outcomes: shorter permit and license cycle times, resident requests resolved without a phone queue, public-safety records and evidence handled faster, budget and procurement cycles that close on time, and a stack that clears the security review the first time. We score seven dimensions: Budget, finance and procurement; Citizen service and engagement; Compliance, security and gov-cloud authorization; Cost economics and time-to-value; Ease of data integration and accuracy; Permitting, licensing and gov operations; and Public safety and justice. The dimensions and benchmarks are public so your council or oversight board can defend the pick, and so vendors can't quietly negotiate them. The audit also captures your operating baselines (agency type, population served, systems of record, and authorization gates) and fans return-on-investment scenarios out per baseline.

3. 03

   We publish. Vendors check facts. Affiliate links are disclosed.

   Every vendor receives their scored tear-sheet seven days before publication and can flag factual errors (wrong pricing tier, misquoted feature, an authorization listed that the vendor does not actually hold, integration listed as native that's actually via a third party). Rankings can't be appealed; only factual corrections are accepted. Where the guide links to a vendor's product, that link may earn us a commission. Disclosed on every page where the link appears. Vendors do not pay for inclusion, placement, or ranking.