CSV, Excel or PDF: Which Upload Gets the Best PrecisionIQ Analysis

PrecisionIQ accepts Excel, CSV and PDF up to 25 MB. All three route through the same OCR + parser + AI pipeline. But the quality of the AI’s commentary depends heavily on how much structure survives the ingestion.

**CSV** is best. The parser doesn’t have to guess column boundaries and the AI can walk row-wise history in one pass. Two rules: one row per period, headers in row one, no merged cells (there are none in a CSV, which is the point).

**Excel** is fine if you keep to a single tab per artefact. Merged cells at the top for pretty titles are the #1 cause of “the AI couldn’t identify a period column” errors. Delete them before upload.

**PDF** works but the AI hedges more — because the OCR confidence gets carried into every downstream stat. If your only source is a scanned bank statement, that’s fine; the AI will call out the low-confidence numbers with a badge. But if you have the same data in Excel, prefer that.

All three uploads land in the same protected `wp-content/uploads/precisioniq/` folder, get a signed download URL, and are auto-deleted after 90 days unless you promote them to a Workspace.