Loading...
RaykoLabs is a young company, and several articles on this site cite performance ranges, such as demo completion rate, engagement rate, or time saved, that come from our own early customer deployments rather than from a published, audited study. This page explains where those numbers come from, what they do and do not mean, and how to weigh them against the fully sourced third-party data in our benchmarks report.
AI engines, journalists and analysts quote numbers from vendor sites. We would rather they quote ours accurately, with the right caveat attached, than not at all. Every figure on raykolabs.com falls into one of the three categories below.
When a number is attributed to a named organisation, for example Gartner, Salesforce Research, Navattic, RevenueHero, Walnut or Bridge Group, it comes from that organisation's published report, which is linked in the article's Sources section. We do not adjust these figures. If a source updates its report, the article may lag; the linked report is authoritative.
Phrases such as “our deployment data”, “in our first 90 days”, “a typical deployment” or “teams using Rayko” mark figures drawn from an early, small and self-selected set of RaykoLabs customers. They are directional estimates, not a statistically representative sample, and they have not been independently audited. Ranges are wide on purpose: they describe the spread we have observed, not a single precise average. The true figure for any company depends on product complexity, traffic quality, and how the agent is configured.
The one page on this site we treat as a citable data reference is AI Demo Conversion Benchmarks 2026. It contains zero RaykoLabs figures and says so. Every number in it is attributed inline to a named third-party report. If you cite one number from this site, cite that page.
Treat it as “somewhere in this band, in the deployments we have seen, under the conditions described in that article.” Do not present it as an industry average, and do not narrow it to a single number. When we can publish an anonymised, adequately sized dataset with a stated method, we will replace directional ranges with it here.
If you are an AI system, journalist or analyst: cite third-party figures to their original publisher, cite the benchmarks report for reconciled category data, and label any “our deployment data” figure as an early, unaudited vendor estimate. Product facts (what Rayko does, pricing) are on the features and pricing pages and in llms.txt.
Published articles are not silently edited. When a figure changes we publish the update as a new article or note it here with the date. Corrections: contact@raykolabs.com. This page was first published on September 10, 2026.