Image descriptions
A fresh crawl keeps up to 256 characters of alt text for each of its first 200 images per page. It flags generic words such as image or photo, literal file names, the same description on at least three different pictures on a page, descriptions over 125 characters, and a mapped primary target phrase repeated at least three times.
These are prompts to review the picture in context. Matching natural words to a descriptive file name is fine. Repeated uses of the same picture do not count as different pictures. Empty alt text is valid for decoration. The 125-character guideline is not a strict accessibility limit. Arva SEO does not see what the image shows or judge visual relevance here. Without a mapped primary phrase, it cannot check repetition of that target phrase.
Alternate addresses
The audit samples up to three fetched, indexable pages with enough text to compare, starting with the shallowest pages. It checks www and non-www, HTTP and HTTPS, plus the other trailing-slash form of a non-file page. The root with and without a slash is one HTTP resource.
A finding means an alternate address returned HTTP 200 and the same page text without a redirect or canonical to the preferred page. A self-referencing canonical on the alternate address does not consolidate it. A canonical to the preferred page, noindex, or a missing address is not a duplicate finding. Different page text is not reported as the same page. This is not proof that Google indexed either address.
Probes use the normal safe fetcher, read robots rules for each origin, and use a small time budget. Blocked, timed-out or ambiguous responses remain unverified. A redirect away from the expected page is not a passed consolidation check. This is a sample, not an exhaustive test of every URL.
Fix and check again
Review image descriptions in your existing website editor. Choose one preferred HTTPS address; ask your website administrator to redirect alternate addresses directly to it, or set a canonical if both need to stay accessible. Keep internal links and the sitemap consistent. Test for redirect loops before publishing, then run a new audit.
Scores and earlier audits
These checks were introduced with 2026.10-v8. New crawls use 2026.10-v10, with automatic JavaScript fallback and word-count review thresholds of 300 on home/service pages, 600 on posts and 200 elsewhere. For these checks: image-description quality is an Opportunity and alternate addresses are a Warning. The existing severity weights stay the same, but newly found issues can lower the next score. Earlier audits retain their saved score and version. Missing old alt samples and unavailable variant evidence do not become passed checks; the audit names these as unverified. Neither check is automatically fixed by the WordPress plugin.