How we make money

Short version: a vendor cannot pay us to change a score, remove a finding, or buy a badge. Here is the whole story.

What vendors cannot do

Re-probes are free, and publish regardless of outcome

Re-checking a listing costs a vendor nothing. If a re-probe finds the flow got worse, the score goes down and that’s what gets published. If it got better, the score goes up. If nothing changed, the old findings stand. We don’t charge for a re-probe, and we don’t sit on results that make a vendor look worse.

Reviews are machines, not people

Every finding on this site is produced by an automated, LLM-driven probe — not a human reviewer with an opinion. Probes run in three escalating-risk tiers (passive/static, unauthenticated interaction, and real signup/checkout attempts gated to vendors that already welcome non-human traffic), and every finding carries the probe tier that detected it and the date it was detected, so the evidence behind a score is always traceable back to a specific probe run rather than an unexplained number. See Methodology for exactly how tiers, timing, and hardness turn into a score.

Our relationship to llm-mart

Interrupt Index shares backend infrastructure with llm-mart, an agent-commerce marketplace project built by the same operator, and it exists partly to feed llm-mart’s own vendor-trust layer — an agent shopping for a tool or API needs to know whether a vendor is actually usable end-to-end before it recommends spending money there. We think that’s worth disclosing plainly rather than letting Interrupt Index read as fully unaffiliated third-party research when it isn’t. What that affiliation does not change: every listing, including any vendor llm-mart might eventually transact with, is scored by the same public rubric against the same probes as every other listing on this site — the guarantees above (no paying to change a score, remove a finding, or buy a badge) apply identically regardless of whether a vendor has any relationship with llm-mart.