How to turn raw property notes into clean listings

If you list property, the bottleneck is rarely finding the information - it's moving it. Viewing notes, an owner's email, a voice-memo transcript: the facts exist, just not in the shape your portal or spreadsheet wants. Here's a repeatable way to close that gap without retyping field by field.

1. Decide your columns once

Before touching any text, define the fields a clean listing needs: address, price, bedrooms, bathrooms, floor area, tenure, EPC rating, and a short description. In Dump2Form these become a template - a reusable schema you set up a single time and apply to every listing after.

2. Paste the mess as-is

Don't pre-clean anything. Paste the raw note, the full email, or the transcript - typos and all. You can attach a photo of a handwritten note or a PDF brochure too; the model reads those directly. One box, no formatting required.

3. Let the AI map, then check the flags

Each field fills in with a confidence indicator and a source. A value pulled straight from your text is marked as data; anything the model looked up online is marked web, with a clickable source and a needs-review flag. Crucially, if a detail isn't present, the field stays blank rather than being invented - so you review a short list of gaps, not the whole record.

4. Export or push downstream

Once the flagged values are confirmed, export to CSV or JSON, or send the record through a signed webhook into your CRM or portal. Web-sourced values are never auto-pushed until you've approved them, so a guess can't reach a live listing on its own.

Why "blanks over guesses" matters here

A wrong square-footage or a fabricated EPC rating is worse than a missing one, because nobody double-checks a field that looks filled. Treating a blank as a valid, honest answer keeps your listings accurate and your review time short.