Audit every image on the site, then fix the ones that matter

Reverse your whole image estate to find what is missing alt text, badly named, or describing nothing, and get the fixes as a diff.

An image audit normally means a spreadsheet somebody fills in by hand and abandons at row sixty. Reversing the estate produces the same spreadsheet with the answers already in it.

Why this will not look generic: Every suggested filename and alt line is derived from the actual pixels of that image, so the audit cannot pad itself with generic fixes that fit any page.

What you walk away with

• Every image on the site, with what it actually depicts
• The ones with missing, duplicated or useless alt text
• A suggested filename and alt line per offender
• A diff you can review rather than a report you have to action

Typical run: 200 to 10,000 across a site. Crawl first and reverse second, so a re-run only pays for images that are new since the last pass.

What the reverse actually gives you

Point it at: Every image on the site, reversed in one sweep.

What comes back (representative example): Across eleven hundred images the picture is lopsided. Four hundred and twelve carry no alt attribute at all. Another two hundred and nine share one of six duplicated strings, almost all of them the company name, which is worse than empty because it reads as deliberate to a crawler. Eighty-one have alt text describing something absent from the picture, usually the heading of the page rather than the photograph underneath it. The rest are fine. Filenames are in worse shape than the alt text: nine hundred and forty are still camera or CMS defaults, and the tags returned alongside each reverse give a usable slug for every one of them.

Things you would never have thought to type:

• 209 sharing six duplicated strings, mostly the company name
• 81 describing the page heading rather than the picture
• duplicated alt being worse than empty, because it looks deliberate
• 940 filenames still camera or CMS defaults

What you change: Nothing by hand. You review a diff rather than fill in a spreadsheet, and every filename suggestion comes from the pixels.

How it works

1. Reverse the estate (In your code) - Point it at your sitemap or your asset directory. Skip anything you reversed on a previous run.
2. Ask which ones actually matter (In Claude) - An audit that flags everything gets ignored. Ask it to rank by page importance and how wrong the current text is.

Cost and time

This workflow runs at 1 reverse + about 1 minute. Your result carries automatically from each step into the next, so there is no re-uploading between tools. Reverses are billed from your reverse meter and generation steps from your AI Credits, both bundled in every pay-as-you-go credit pack.

Browse all guided workflows or all individual tools. Pay as you go, no subscription, AI Credits never expire.