# Bookkeeper · Day 1 — Try it on this messy bank export first

**Subject:** Try it on this messy bank export first
**Send:** Day 1, ~9am buyer-local-time
**Goal:** convert "I bought it" → "I ran it on something"

---

Hi {{first_name}},

Yesterday's email had your download. Today's email has a *file* — a sample bank export I built specifically to break things.

→ **{{sample_file_url}}** (260 KB CSV, 1,400 rows of synthetic data — no real account info)

It's modeled after real exports I've seen from US, UK, and Canadian banks. Hidden in there:

- Mixed date formats (some `MM/DD/YYYY`, some `DD-MM-YY`, one row in `YYYY-MM-DD`)
- Six different spellings of "Amazon" across the merchant column
- Trailing whitespace + non-breaking spaces in the description column
- Three obvious duplicate transactions and two non-obvious ones (different timestamps, same amount + merchant)
- A totals row at the bottom that's not a transaction
- One row with currency in `€` instead of `$`

Drop it into DataTools, click **"Run all"** in the analyzer, and look at the gate report. It'll catch all of the above and tell you exactly what changed and why.

The audit trail (a sidecar CSV called `<filename>.audit.csv`) is the part most bookkeepers are surprised by. Open it in Excel — every change has a row: original value, new value, rule that fired, timestamp. That's the file you hand to your client when they ask "wait, why did you re-classify that?".

Try it once on the sample, then once on a real client export. Reply and tell me what it caught (or missed) — I'm building the v1.1 detector list from real-world feedback.

— Michael
{{support_email}}