Methodology
How this report was built: data sources, scoring, profiling, and what the numbers mean.
Two Layers of Analysis
This report combines two distinct layers of analysis. Understanding which layer produced which claim is essential for evaluating the intelligence.
Layer 1: Starbridge Budget Intelligence (the scoring)
Who did this: Starbridge (starbridge.ai) — an independent budget intelligence platform.
What they analyzed: 7,186 US cities. For each city, Starbridge reviewed:
- Adopted budgets and multi-year financial plans
- Capital improvement programs and bond activity
- Leadership changes (mayors, city managers, CIOs)
- Program launches, hiring signals, and fiscal news
- Budget source documents (avg. 7.2 source URLs per city)
What they produced: A Budget Trend Score (0-100) and a Propensity to Spend rating for each city, plus an AI-generated budget analysis summary citing specific fiscal evidence.
How the score works: Cities at the top (80-100) show expanding budgets, new capital programs, and active procurement signals. Cities at the bottom (5-30) show tightening spend, hiring freezes, and constrained fiscal conditions. The median score is 51.
Layer 2: Content Intelligence Pipeline (the profiling)
Who did this: This pipeline — built as a proof of concept for Starbridge.
What we did:
- Parsed all 7,186 city records from Starbridge's dataset into a queryable database
- Enriched 12 cities with public web research (budget documents, news articles, city websites) — adding total budget figures, named capital projects with costs, and leadership verification
- Generated 52 city intelligence profiles with sourced claims and inline hyperlinks
- Clustered cities by geography, investment theme, and spending behavior to produce cross-city market intelligence articles
What we did NOT do: We did not create the scores, compute the propensity ratings, or write the budget analysis summaries. Those are Starbridge's work. We transformed their structured data into publishable content.
Full Dataset Distribution
These numbers come from all 7,186 cities in Starbridge's dataset — not just the 52 profiled in this report.
How to read these numbers
Every number below comes from Starbridge's dataset — specifically, the Propensity to Spend and Budget Trend Score columns in their CSV export of 7,190 city records. We loaded 7,186 of those records into our database (4 rows lacked a state code and were excluded). We did not compute these classifications. Starbridge assigned them. We report what they assigned.
Propensity Distribution (Starbridge's Classification)
Propensity to Spend column equals "Aggressive Spender." We mapped this label to "aggressive" in our database. Starbridge defines the threshold; we report the count.Score Distribution (from Starbridge's Budget Trend Score column)
Starbridge assigns each city an integer score. In this dataset, scores range from 5 to 92 (not 0-100 — whether 0 and 100 are achievable is defined by Starbridge's methodology, not ours). Median score: 50. Mean: 50.9.
We grouped scores into three momentum bands for navigation. These thresholds are ours, not Starbridge's:
- Accelerating (score 70+): 1,242 cities (17.3%) — we chose 70 because it's near the 75th percentile (P75 = 68)
- Steady (score 40-69): 3,289 cities (45.8%)
- Tightening (score below 40): 2,655 cities (37.0%)
About the 52-City Sample
Selection criteria (our choices, not Starbridge's)
We selected cities where Budget Trend Score >= 72 AND Population >= 10,000. This intentionally skews toward high-score, mid-to-large cities — the ones with the richest Starbridge summaries and most web-searchable budget data. It is NOT a representative sample of all 7,186 cities.
What this proves: The pipeline works. Given Starbridge's data, we produce publishable intelligence profiles. What this does NOT prove: quality at the low end of the score range (cities scoring 5-40) or for very small municipalities (<5,000 population).
Two tiers of profiles
12 enriched profiles combine Starbridge data with public web research. For these cities, we ran web searches for budget documents and news articles, extracted total budget figures, named capital projects with costs, and verified leadership names. Every enriched fact is an inline hyperlink to its source — click any purple link to verify. Example: Oak Harbor, WA (enriched).
40 standard profiles use Starbridge's dataset only — no web enrichment. These are thinner but honest. Every claim traces to Starbridge's Propensity to Spend Summary field. Example: Woodbury, MN (standard).
With full API access to Starbridge's gated columns (operating budget amounts, IT spending, procurement navigation guides, fiscal year dates), every city would get the enriched treatment — sourced from Starbridge's own data rather than public web research.
Attribution Standard
Every factual claim in this report follows one of two attribution paths:
- Starbridge data: Budget Trend Scores, propensity ratings, and budget analysis summaries are attributed to "Starbridge budget intelligence data, December 2025." These are Starbridge's assessments, not ours.
- Public sources: Enriched facts (total budget figures, named capital projects, leadership names) are inline-hyperlinked to their source — a news article, budget document, or city website page. Click any purple link to verify the claim at its source.
If a fact cannot be attributed to either Starbridge data or a verifiable public URL, it is omitted. No unsourced claims appear in any profile.
Scale This to Your Full Dataset
This report profiles 52 cities. The pipeline processes all 7,186. With Starbridge's full API access — including operating budgets, IT spending, and procurement guides — every city gets the enriched treatment shown in the top profiles.
Schedule a strategy session →