What can do?

A score in about 20 seconds. Enter an address and get a 0–100 score, a verdict and every check, with no signup.

82 checks on every page. 48 scored checks across 7 courses, plus 20 in Advanced and 14 in Good practice that we measure without scoring.

Read the way machines read. We fetch the raw HTML the way most AI crawlers do, and open the page in a real browser only when the HTML looks thin, so you see what only appears after scripts run.

The site’s own files, every time. Every page scan also reads robots.txt, the sitemap, llms.txt, security.txt, the favicon, the /.well-known/ agent files and an AI-discovery DNS lookup.

Redirects followed and named. If the address redirects, we scan where it lands and tell you so.

Performance, on its own dial. Google PageSpeed’s lab score and 28 days of real-visitor data, shown beside the score and never inside it.

Your structured data, shown back to you. The JSON-LD on your page, laid out to read, with a copy button.

Scan it again. A fresh scan whenever you like; the old report keeps its own link.

A verdict in plain English. From Chef’s kiss to Sent back to the kitchen, and never better than how well AI can read the page.

The tasting menu. A score for each of the 7 scored courses, then every check, grouped by course.

Measured, not scored. Performance, Advanced and Good practice are shown in full and never move your score.

Four answers, never a wall of red. Pass, Worth a look, To fix and Couldn’t judge, each with its colour, glyph and word.

Every finding explained. What we found, why it matters, what good looks like, and a page for each check.

“Couldn’t judge” is an answer. If we couldn’t look, we say so and leave it out of the score. A page that refuses us gets no score, never a zero.

Told when we’re blocked, and how to let us in. We say how we were stopped; when robots.txt refused us, you get the two lines that let our bot in.

Errors that say whose side they’re on. In your browser, on our side, on the site’s side, or this link.

A permanent link for every scan. It opens that report exactly as it was on the day, performance included.

History that builds itself. Every scan of that exact address, whoever ran it, with the score change between each one.

Our changes, marked as ours. When our scoring changes, the number still shows, marked “our scoring changed”, with a link to what changed.

Compare any two scans. Tick two scans for the scores side by side, every course that moved and every check that changed.

Other pages you’ve scanned on a site. Your other scans on the same site, listed on the report, in this browser only and never shown to anyone else.

Up to 50 pages, chosen by kind of page. Enough of each template to stand for the rest, and every page gets the same full scan.

Three ways to say what to scan. Your domain (we find the sitemap), a sitemap address, or a pasted list of up to 2,000 addresses on one site; we read the first 50, in your order.

The shape of your site. One row per kind of page, with the range and median of their scores.

A site score that isn’t an average. Each check is judged across every page we read, with credit for the checks that are mostly right.

Findings with their reach. Each finding says how many pages it is on, with links to those pages’ reports.

What we could and couldn’t judge. How many pages we read, how many we could judge, and which we set aside and why.

Pages robots.txt keeps from us. Named, with the rule that matched them.

Homepage performance. PageSpeed runs on the homepage of every site scan, shown apart from the score.

Leave and come back. Live progress while it runs, its own link, and an email when it’s done.

Your sites. Every site you scan in one place, with the latest score, what needs a look and a sparkline for each.

One page per site. Score over time, where the score comes from, what changed and every run.

What changed since last time. Checks fixed and checks that started failing, course by course, on the pages both runs read.

Pin the pages that matter. Up to 25 pages read in every run, on top of the sample, like your pricing page.

Compare two site scans. Two runs side by side; if the pinned pages changed between them, we say so instead of subtracting.

Monthly re-scans. Every site is scanned again 30 days after its last scan, with nothing to set up. Switch it off any time.

The result by email. One email when a site scan finishes and one after each monthly re-scan, with the new score beside the last.

Yours to remove. Removing a site stops its re-scans; deleting your account erases your email, sites and pins. Page results stay with the page.

Send any report. Every page scan has a link anyone can open, with no signup, and it doesn’t expire.

A preview card when you paste it. The link unfurls as a card with the score, verdict, page and date.

Share a site scan. Private until you create a share link; then anyone with it reads the whole report, every page and every finding, without an account.

Stop sharing. One button, and the old link stops working straight away.

Opened from a link? You see the whole report, nothing is asked of you, and grey never counts against the site.

A page for every check. 82 pages: what the check looks for, why it matters to a machine and what good looks like.

9 course hubs. One page per course, with its checks and how they fit together.

Glossary. 221 terms, from AEO to Web Bot Auth, in plain English.

AI crawler directory. 25 AI crawlers: who runs each, what it’s for and whether it follows robots.txt.

The Webflow SEO & AEO checklist. 52 items in 7 phases, with your progress saved in this browser.

Help. How-to guides for every part of Bot Appétit.

The share of sites passing each check. Refreshed every hour, percentages only, and a figure appears only when enough sites sit behind it.

Where sites differ. By platform, by hosting, and homepages against inner pages.

Download and cite. The figures as CSV and JSON under CC BY 4.0, with how to cite them.

Share a figure. An image, an embed and share text for each headline figure, each with a copy button.

A named, signed bot. BotAppetitBot signs every request it makes (Web Bot Auth, RFC 9421) with a key we publish.

Our bot page. What we fetch, what we keep, and how to let us in or keep us out.

A public changelog of how we judge. Every change in plain English, marked when it may move scores.

A real studio behind it. When you’d rather it were simply handled, Milk Moon Studio handles it. Book a call: no pitch, just a plain-English chat.