What a person decides
The judgment calls that are yours to make, and where you make them in CampusMind.
Remediation is a division of labour. Automation rebuilds structure across hundreds of documents and proves it passed; you bring the meaning — whether an image description says the thing the image is there to say. Both halves matter, and CampusMind is designed so each is done by whichever is better at it.
Who settles what
| The machine settles this | You settle this |
|---|---|
| Is there a tag tree, and is it well formed? | Does the heading structure match the document's actual argument? |
| Does every image carry alt text? | Does that alt text convey what the image is for? |
| Do tables have header cells with a scope? | Do those headers describe the right cells? |
| Is a reading order defined? | Is it the order a person would read aloud? |
| Is the document language declared? | Is it the right language? |
The three that need you most
Generated descriptions are literal by design — they describe what is visibly present. That is exactly right for a logo, and the starting point for a chart.
- Alt text on anything that carries an argument. "A line chart with a downward slope" describes the picture. "Enrollment fell 12% between 2023 and 2025" describes the point. Only you know which one the document needs.
- Reading order on complex layouts. Multi-column pages, pull quotes between columns, and tables that break across a page are where automatic ordering is most likely to be wrong. Worth a look every time.
- Contrast. A contrast failure lives in the source design. Preserve Layout reports it and leaves it alone, because silently restyling a document would misrepresent it.
Where you do this
Edits happen in the Tag & Caption Editor. Changes are held locally until you save, so you can work through several and commit them together. Saving creates a new version, and Revalidate re-runs the PDF/UA-1 check against your edited file.
How review works
CampusMind uses a review-and-edit model: remediated documents are ready to use the moment they finish, and you improve the ones worth improving without waiting on anyone. Institutions that run a formal review stage keep it in their own process and use version history as the record of what CampusMind changed and when.
Why this matters beyond usability
Automated processing resolves roughly 80% of common accessibility issues — the structural work that would otherwise take a specialist hours per document, done in minutes. The remaining share is small in count and large in effect: it is disproportionately the part that decides whether a document genuinely reads well.
That is why the editor exists, and why regulators expect a person in the loop — see the Department of Justice discussion referenced on ADA Title II. The model puts your attention on the 20% that needs it, instead of the 100% that used to.
Related
- Accessibility standards — what the score covers.
- Version history — what the record shows.
- How CampusMind works — the pipeline behind the settings.