Spent the last few years building tools for freelance investigative journalists, and the same pattern shows up over and over: reporters find a great public dataset, spend a weekend fighting Excel, and either miss the story or publish something with a fixable error in it. A few things I've learned watching this happen:

1. Government CSVs are never clean. Column headers change mid-file, dates come in 3 different formats, and missing values get encoded as blank, "N/A", "-", and "999" — sometimes all in the same column. If you don't standardize this first, your averages and totals will just be wrong, and you won't know it.

2. Duplicate rows hide in plain sight. Agencies re-publish updated versions of the same record without removing the old one. If you're counting rows instead of deduplicating by a unique ID, your "55% increase" headline might actually be a reporting artifact.

3. Joining two datasets by name instead of ID. Matching "Robert Smith" across a court records file and a campaign finance file feels fine until you find three Robert Smiths in the same county. Always look for a stable ID field before you join.

None of this requires a data team — it just requires slowing down for 20 minutes before you start charting. That's the exact gap we built Vantage to close: source, clean, and visualize public data without needing a SQL background or a newsroom budget. If you're a freelancer working with public records, would genuinely love to hear what tools/hacks you use for this.