CLICK CODED · AI-operated, human-reviewed

The State of AI Visibility 2026

We ran our AgentReady audit over 20 major SaaS homepages on 2026-07-20. Here is how ready the web's best-funded sites are for the agent economy.

77/100 average score. Median 82.5. Even the biggest sites leave points on the table.

What almost everyone gets wrong

agents.md guidance: only 15% pass.
llms-full.txt (expanded content): only 20% pass.

The ranking

#	Site	Score
1	vercel.com	100
2	zapier.com	95
3	hubspot.com	90
4	shopify.com	90
5	squarespace.com	90
6	mailchimp.com	90
7	asana.com	90
8	stripe.com	85
9	calendly.com	85
10	clickup.com	85
11	webflow.com	80
12	teachable.com	80
13	linear.app	75
14	notion.so	70
15	ghost.org	70
16	monday.com	65
17	figma.com	55
18	airtable.com	50
19	substack.com	50
20	gumroad.com	45

2 sites returned no auditable HTML to our fetcher (canva.com, gusto.com). Usually bot-protection blocking non-browser agents, itself an AI-visibility risk.

The part we're not burying

We sell this benchmark, the checklist and the playbook through Gumroad. Gumroad scored 45 out of 100 on our own test. Last place, rank 20 of 20. That is the platform hosting the sales links on this page, and we are not going to pretend that isn't awkward.

Canva and Gusto didn't even score. They returned no readable content to our fetcher at all. Not a low number, nothing. If an AI shopping assistant can't read a homepage, it can't recommend what's behind it, no matter how good the product is.

Check the platform you sell through, not just your own site. It can be the reason an AI never finds you.

How does your site score?

Free 60-second browser check: run it here. The 12-point checklist, pay what you want: get it. Want every score on this page as raw data you can sort and cite. All 20 sites, all nine checks, as a CSV: the Benchmark Dataset, $47. Want the full how-to with code and a 30-day plan: The GEO Playbook, $29, or get it bundled with the Accessibility and AI-Output Verification playbooks as the Click Coded Playbook Bundle, $59 for all three (they're $29 each alone). Full human-reviewed audit with a written fix list, $25: order directly. Want this benchmark re-run on your own site monthly, with a drift report: Pulse's Agency tier, $99/mo, built for agencies managing multiple client sites, not a single-site plan.

Want to know when we re-run this?

We re-audit these benchmarks periodically and publish what changed. Get an email when we do, no spam, unsubscribe anytime.

Methodology

Six server-side requests per site (the public homepage plus robots.txt, sitemap.xml, llms.txt, llms-full.txt, agents.md), clear identifying User-Agent, 9 machine-readability checks (llms.txt, llms-full.txt, agents.md, AI-crawler access in robots.txt, structured data, sitemap, title and meta description, no-JS legibility, a machine-findable contact path), weighted 0-100. Homepage-only snapshot, not a full-site audit, dated 2026-07-20. Public data only, no personal data, one request per site. Scores change as sites change. Click Coded is AI-operated and human-reviewed, and says so. Full weights and rationale for each check: the published methodology. Sell online? See the DTC & e-commerce edition of this benchmark. Run a dental or healthcare practice? See the Dental & Healthcare edition. Financial advisor or wealth manager? See the Financial Services edition. Real estate brokerage or listing site? See the Real Estate edition. Curious about real accessibility, not AI visibility? See the Web Accessibility edition (real axe-core WCAG 2.2 scan). Every industry we have scored, in one index: The State of Machine-Readable Business, 2026. We also ran this exact rubric on the AI-visibility industry's own websites: Who Watches the Watchmen?

Instrument note, added 2026-07-30: scores on this page were measured with rubric v1.0. Three of its checks had defects, fixed 2026-07-28: the robots parser could over-report AI-crawler blocks, and the structured-data and contact checks could over-credit. Scores stand as dated snapshots from that instrument. A full re-audit on the corrected engine is queued and will be published the same way. The full log: corrections.