Vendor Check: did the agency you hired actually move anything
Pick the date a vendor or campaign started and test whether your measured AI visibility genuinely changed since.
What it is
A before-and-after test on your own numbers. You give it the date the work began, and it compares your last measured run before that date against your latest run since, using the same significance test we apply to every other change on the platform.
When to use it
When you are paying an agency, a freelancer, or an AI-SEO package and want to know whether anything actually happened, rather than reading a report written by the person you are paying.
Step by step
- 1Open Vendor Check from the sidebar.
- 2Enter the date the work started, and optionally the vendor name for the report header.
- 3Click Check the period.
- 4Read the verdict, the two measured points, and the confidence interval around the change.
What you will see
A before figure and a latest figure with the prompt count behind each, the change in points, a 95 percent interval, and the p value. It also reports how much your baseline runs varied before the work started, so a volatile starting point is visible instead of hidden inside a single number.
Measured, never manufactured
This answers whether your numbers moved, never who moved them. There is no verdict state meaning the vendor worked, and the attribution caveat renders beside every result, because a date split cannot establish cause. Seasonality, your own shipping, and an engine changing its behaviour all live in the same window. A change that falls inside the noise is reported as no measured change rather than dressed up as a small win.
Related guides
Next
How we measured thisA standing per-pillar checklist of the exact checks behind each score, on Analyze, Monitor, Optimize and the shareable report.