Databook

This update introduces dedicated public pages for software license families, allowing users to view merged spellings, stats, and contracts. It also links vendors on these pages directly to their profiles when there is an unambiguous match, and displays the raw spellings combined into each family for transparency.

This update improves the transparency of our license contract data by disclosing when contracts contain duplicate rows with conflicting information. The API and public-facing methodology notes now clarify how these ambiguities are resolved, and the data tables now explicitly state when contract values and procurement routes are indicative rather than exact.

Updated the Licenses page to replace a misleading "100% not competitively bid" metric with a detailed breakdown of actual procurement routes. The page now highlights the most common route (57% small purchases) and provides a clear table explaining what each route means to ensure accurate interpretation of city purchasing data.

This update introduces an unlisted Software Licenses research page that groups and analyzes 948 AI-detected city software contracts worth $1.37 billion. The page provides five analytical views—including vendor concentration, renewal calendars, and contract fragmentation—with support for CSV exports and deep links to renewal queues.

This update fixes a display bug on the expiring contracts page, ensuring the "Build-vs-buy" dropdown filter renders clean, readable text instead of raw HTML code. Additionally, the digital services contract classifier now automatically calculates and reports its API token usage and financial spend after each run.

This update fixes a bug on the district pages where certain district types, such as Neighborhood Tabulation Areas (NTAs), would fail to load their capital project statistics. The system now gracefully handles missing district data and displays a dash instead of causing a page error.