Check image alt text coverage

This checker reviews each record without fetching the image or page, calculates the percentage that passes, and returns every offender with its position, source, original alt value when present, and failure reason. It is designed for repeatable SEO and accessibility quality checks in publishing workflows, site crawls, content migrations, and release gates.

What the coverage percentage tells you

A simple count of existing alt attributes can make an image library look healthier than it is. An attribute containing only spaces, a label such as “image,” or a camera filename does not communicate useful subject matter. This checker therefore counts an image as covered only when its alt value is non-empty and does not match a conservative list of generic placeholders. The percentage is the number of passing images divided by the total number supplied, rounded to two decimal places. Alongside that headline result, the covered and total counts make the calculation transparent. The offenders array identifies every record that reduced the score, preserving the zero-based input position and the supplied src value so it can be traced back to a crawler export, content database, or template. This is a coverage signal, not a judgment of writing quality: a specific but inaccurate description can pass because the checker does not inspect image pixels or surrounding page context.

How missing, empty, and generic values are classified

Each offender receives one of three reasons. Missing means the record has no alt property at all, which often indicates that the HTML attribute was absent. Empty means an alt property exists but contains no characters after surrounding whitespace is removed. That distinction matters because an intentionally empty alt attribute can be correct for a decorative image, even though it is still excluded from this coverage metric. Generic means the value is present but matches a deliberately narrow placeholder rule. Examples include “photo,” “picture,” “thumbnail,” numbered labels such as “image 2,” and common camera-style filenames such as IMG_2048.jpg. Matching ignores case, repeated spaces, underscores, hyphens, and trailing punctuation where appropriate. The checker does not reject short text merely for being short and does not perform fuzzy semantic guessing. This conservative behavior keeps results deterministic, reviewable, and suitable for automated release thresholds without unexpectedly condemning legitimate product names or concise icon descriptions.

Use the offender list in an SEO workflow

Supply records from a crawler, CMS export, component inventory, or test fixture. Every record needs a non-empty src identifier; it may be a full URL, a relative path, or an internal asset key. The alt field is optional so that a missing HTML attribute can be represented honestly. After running the check, sort remediation work by the returned reason. Missing attributes usually point to markup or component defaults, generic values often require editorial replacement, and empty values need human review to separate purposeful decorative images from overlooked content images. Run the same list again after corrections and compare covered, total, and percentage values in a build report or audit trail. Because the algorithm uses no network, image analysis, random values, or current time, identical input always produces identical output. Requests with an empty image list are rejected rather than reported as zero percent, since a percentage over no observations would be misleading and could let a broken crawler appear to complete a valid audit.