Metra first checks each PDF page for usable native text. Pages containing a scanned table are queued for optical character recognition (OCR); pages with an adequate text layer keep their original text. The uploaded PDF remains the source document. OCR is a machine interpretation and does not alter it.
Wait for the OCR attempt to finish before mapping. Progress is page based and is not an exact completion-time estimate. You can cancel queued or running recognition. If a safe processing failure occurs, correct the scan or use Retry where available.
queued means the recognition request is saved and waiting to start. It does not confirm that any pages have been processed. Delayed starts and interrupted processing are recovered automatically; you can leave and return to this import to check the same attempt. Do not queue another attempt while the saved one is active. If the wait is unusually long, contact support with the import reference, attempt number and request time. Completion time varies with the document and workload.
Resolve uncertain values
Review every marked asset ID, serial number, customer, site and location value against the source page. For each uncertain token, either accept the recognised value, enter a correction, or exclude it. Corrections and acceptance are attributable review decisions; Metra does not silently guess confusing characters such as 0/O or 1/I.
An unresolved identifier prevents final validation. After review, continue through the normal table selection, mapping and validation steps. No equipment or location records are created by OCR itself.
For best results, upload a straight, high-contrast scan at about 300 DPI. Encrypted PDFs, handwriting, severe perspective distortion and photographs are not supported.