Parsing Issues

CraftStage's AI parsing handles most data formats automatically, but some inputs can be tricky. This page covers common parsing issues and how to resolve them.

Low confidence scores

Symptom: Many items are flagged with yellow or red indicators.

Common causes:

• Ambiguous formatting. Mixed formats in the same list (some "First Last", some "Last, First") confuse the parser.
• Missing context. A bare list of names with no headers or labels gives the AI less to work with.
• Non-standard separators. Using pipes (|), semicolons, or unusual delimiters instead of commas or line breaks.

How to fix:

• Review each flagged item and correct it manually.
• For future orders, ask customers to use consistent formatting.
• Add a header line to your paste (e.g., "Name - Title") to give the AI context about the format.

Names appearing garbled or with wrong characters

Symptom: Characters like "José" instead of "José", or "O’Brien" instead of "O'Brien".

Common causes:

• Encoding issues. The source text was copied from a system with different character encoding (e.g., Windows-1252 vs. UTF-8).
• Smart quotes. Word processors often convert straight quotes to curly quotes, which can get mangled during copy-paste.

How to fix:

• Edit the affected items manually to correct the characters.
• If the issue is widespread, try uploading an Excel file instead of pasting text. Excel files preserve character encoding more reliably.
• When copying from emails, try copying from the plain text version rather than the HTML-formatted version.

Excel file not parsing correctly

Symptom: Columns are misaligned, data appears in wrong fields, or rows are missing.

Common causes:

• Merged cells. Excel files with merged cells can confuse the parser.
• Multiple sheets. CraftStage reads the first sheet by default.
• Hidden rows or columns. Data in hidden areas may or may not be included.
• No header row. Without headers, the parser can't identify columns.

How to fix:

• Clean up the Excel file before uploading: unmerge cells, remove blank rows, and make sure headers are in the first row.
• If data is on a different sheet, move it to the first sheet or copy it into a new file.
• Save as CSV if Excel formatting is causing issues — CSV is simpler and avoids most formatting problems.

Name reordering mistakes

Symptom: "Smith, John" was parsed as "John Smith" but the original format was intentional (e.g., alphabetical listing).

What's happening: The AI detected "Last, First" format and reordered it. This is usually correct for personalization but not always.

How to fix:

• Click the yellow flag on the affected items and edit them back to the intended format.
• If all names should keep their original order, edit them in batch on the job detail page.

Parsing fails entirely

Symptom: "Couldn't extract names" error message.

Common causes:

• The pasted text is in a format the AI can't interpret (e.g., a table rendered as spaces, an image of text, or heavily formatted HTML).
• The input is very short or contains no identifiable name patterns.

How to fix:

• Try reformatting the input as a simple numbered or line-separated list.
• If pasting from a website or formatted document, try pasting as plain text first (Ctrl+Shift+V in most applications).
• As a fallback, add items manually using the "Add Item" button on the job detail page.