Recruitment technology
CV parsing and OCR: fix document quality before AI screening
A CV comparison can only assess the information that reaches it. A scanned page, unusual column order or missing text layer can change what the system receives. Treat extraction as a distinct quality step, with its own checks and exceptions, before asking a reviewer to interpret the resulting assessment.
Compare the extracted text with the original
Use a small sample covering a simple document, a multi-column CV and a scanned file. Check names, role dates, qualifications and the order of experience entries. Look for a qualification detached from its provider or a table read across the wrong columns. Record the source document and parser version so a correction can be traced to the application that was actually assessed.
Distinguish a technical failure from an evidence gap
Create separate statuses for unsupported format, extraction needs attention and ready for review. If text is absent because a page was not read, avoid reporting that the applicant lacks the relevant skill. Route the file to an owner who can inspect it or request an accessible alternative. Make the exception visible in batch totals so failed files are not silently dropped from the shortlist.
Request alternatives without penalising applicants
Use clear instructions for accepted formats and offer a contact route where the upload does not work. W3C guidance explains the importance of associated form labels. Apply that principle to upload controls and error messages so applicants understand the required action. Where an adjustment is needed, use the organisation’s appropriate process rather than asking the person to work around an inaccessible interface unaided.
Retest extraction changes before scaling
Keep a fictional test set containing layouts that previously caused problems. Rerun it after changing a parser, OCR setting or document conversion library. Record which errors remain and whether the review team can identify them. Do not describe OCR as perfect or assume a higher character count proves accuracy; a small amount of misplaced text can matter more than a large amount of correctly extracted boilerplate.
Sources and further reading
Sources checked 25 October 2026. The workflow recommendations are MyValidCV editorial guidance. Check current requirements with the relevant authority; this article does not establish legal compliance or certify a product.
Explore a human-reviewed CV workflow
MyValidCV Enterprise supports bulk CV-to-job comparison, evidence review and applicant email drafts for review before sending. A comparison score is advisory; your team owns hiring decisions. The operational practices described here may require separate systems and policies.
Connect your HR software
The MyValidCV API pilot brings CV-to-job comparison into your own recruitment workflow, with owner-approved company access, capped usage and human review. Request a company pilot.