Scan preparation
Prepare supplier scans before converting tables to Excel
Improve the input, preserve evidence, and test cleanup changes without erasing faint characters or table details.
Preserve the original before cleanup
A poor scan can make a table difficult to read before any conversion begins. Keep the received file unchanged and create a working copy for adjustments. Record missing pages, cropped edges, or unreadable characters at intake. Software cleanup cannot reliably recover information that is absent from the source.
If the supplier can resend a clearer copy or an original spreadsheet, ask through the normal business process. Otherwise, identify the pages that need attention rather than applying a heavy filter to an entire batch by default. Clear pages provide a useful control when evaluating whether an adjustment helps.
Inspect the details that matter
Look at SKU characters, decimal marks, minus signs, thin column boundaries, and the smallest printed text. Also inspect shadows, skew, perspective, and any cropped table edges. A page can look pleasant at normal viewing size while still making a critical digit ambiguous.
Adobe documents scan settings including deskew, background removal, resolution, and text sharpening. These are available adjustment categories, not proof that a particular setting improves your table. Choose a small difficult area, keep its original version, and change one setting at a time so the effect can be understood.
Avoid cleanup that removes evidence
Check whether a proposed filter erases faint punctuation or makes neighboring characters merge. Compare the cleaned version with the original before proceeding. A stronger contrast setting may improve one region while damaging another, so inspect both light and dark portions of the page.
Do not repaint a doubtful character to match what seems likely. If a value requires interpretation, keep that uncertainty in the extraction review rather than altering the source image. For a recurring process, record which preparation steps were applied so that an unexpected result can be reproduced and investigated.
Test a small conversion after each change
Use the same output fields and acceptance checks for the original and adjusted sample. Compare complete rows, especially identifiers and amounts. Record whether cleanup affects row grouping as well as text recognition. Better-looking text does not necessarily mean that the table structure becomes more accurate.
For a single photographed table, Microsoft's Data from Picture instructions recommend a focused, head-on image containing the relevant table. It can be an already available evaluation route for eligible Excel users. Treat it as a separate workflow with its own review step, not as evidence that all PDF batches can use the same method.
Build a repeatable preparation rule
After a successful small trial, apply the chosen steps to a modest batch from the same source type. Keep exceptions separate when a page has different quality or layout. A stable rule should specify when it applies and when an operator should stop and inspect.
Retain the original, prepared copy, and extracted result with clear names. Record unresolved input problems alongside the output instead of allowing them to become unexplained spreadsheet errors. Preparation is successful when it helps produce verifiable data while preserving a clear route back to what the supplier actually supplied.
Sources
Evidence status: methodology. No unverified accuracy, savings or traffic claim is made.