WorkBuddy field misalignment when merging industry tables - am I using it wrong?

Last night, my advisor asked me to add a sentence about the industry background in Chapter 3. I came across sales data for aerial work platforms for Jan-Feb: total volume of 24,250 units, up 9.89% year-on-year. So I casually tried to extract the domestic, export, and YoY columns from several PDFs and Excel files to make a small table.

I've been using WorkBuddy for about a month, usually just letting it organize literature packages. I previously wrote a post about using it for my thesis pipeline, thinking that as long as I didn't let it write the main text, it would be safe. But this time, I couldn't even get the table right. I created an "Industry Data" folder, dropped in three PDFs, two screenshots, and one Excel file, and asked it to extract by "Month, Category, Sales Volume, YoY." I specifically told it not to generate body text and not to touch the thesis content. The resulting table had misaligned fields: 24,250 ended up in the "Export" column, and 9.89% sometimes appeared in "YoY" and sometimes in "Notes." Even weirder, it treated repeated headers as data, generating a bunch of empty rows labeled "YoY decline."

I tried renaming the files and also tried uploading only one PDF at a time, asking it to output field mappings first before merging. But the second attempt prompted a lack of permission to access local folders, and when exporting to CSV, a column was missing. Maybe I misunderstood something—is WorkBuddy unable to directly process these statistical bulletin-style tables? Do they need to be converted into structured Excel first? Has anyone else encountered this? Did I mess up some setting?