Two different questions wearing one label
Tools in this market answer one of two questions. The first is visibility: when someone asks an AI assistant a question, does your brand appear in the answer? Profound, Peec and Otterly live here, alongside modules bolted onto Semrush, Ahrefs and HubSpot. The second is usability: when an agent arrives at your site with a task, can it complete the task? Cloudflare's checker, Ora, Lighthouse's agentic browsing category, Proven's AgentFriendly ranking and SecondDoor live here.
The two get conflated because both say AI somewhere on the page. They fail differently: a brand can lead every AI answer and still lose the purchase at a checkout the agent cannot pass, and a perfectly agent-usable site nobody cites gets no agent visitors to use it. Serious teams end up needing an answer to each question, and no single tool measures both well as of August 2026.
The visibility monitors
Profound is the enterprise standard: around 700 enterprise customers, 100 million AI queries tracked a month, and a Series C at a billion-dollar valuation in February 2026. It starts at $99 a month and climbs quickly. Peec is the mid-market pick: unlimited seats, agency workspaces, roughly 85 euros to 505 dollars a month, and the fastest growth in the category. Otterly is the budget entry at $29 a month with a static readiness audit bundled in.
All three monitor and benchmark. None walks your site, and the most common complaint in their reviews is that they report positions without changing them. Buy them to know where you stand in answers, not to fix your funnel.
The usability scanners, all free so far
Cloudflare's Agent Readiness checker scores your protocol surface, bot access and capability endpoints, with a fix prompt per failure. It is authoritative on edge and bot questions because Cloudflare operates the edge, and it states plainly that it does not walk a funnel. Ora, distributed by Vercel as is-agentic.com since August 2026, runs 127 static checks plus Journeys, which gives an agent a real task and narrates where it struggled; its public leaderboard passed fifty thousand sites. Lighthouse 13.3 added four agentic browsing checks inside Chrome itself, the strongest sign this is becoming a mainstream discipline. Proven ranks 510 plus B2B SaaS vendors on their API surface. A long tail of single-page checkers copies the same static checks.
Everything in this paragraph is free, and everything in this paragraph looks at one door: the agent's. None of them measures whether agents are treated worse than the people your site was built for, because none of them establishes what a person experiences on the same pages.
Where SecondDoor sits, stated plainly
SecondDoor scans the site twice on every scan, once as a human visitor and once as an AI agent, in a real browser, and scores both funnels out of 100. The difference is the Door Gap, and it is the number this product exists to measure. Eight of ten dimension scores are computed deterministically from fetched evidence, unmeasured is reported as unmeasured rather than guessed, and deep scans keep screenshots of the walk. The basic scan is free, public and permanent; paid tiers from $29 a month add scheduled rescans and regression alerts.
SecondDoor does not track citations, and does not intend to. If your question is visibility, use one of the monitors above, and several of them are excellent. If your question is whether the agents those tools count can actually buy from you, sign up, or read your pricing, that is the door this tool was built to test.