Source Trust: which sources AI engines are citing less over time

The direction of each cited source's share, read across every answer we have stored rather than from a single lookup.

What it is

A list of the domains AI engines cite, with whether each one is gaining or losing share of citations over time. It reads every stored engine answer across every site we sample, groups cited domains by week, and reports the direction.

When to use it

Before you spend money or effort on getting mentioned somewhere. A source quietly losing citation share is one to stop investing in, and that is a decision you cannot make from a snapshot.

Step by step

1Open Source Trust from the sidebar.
2Read the top of the list, which is sorted steepest decline first.
3Check the citation count beside each row before trusting its direction.
4Use it as a spending filter, not as a to-do list.

What you will see

Domains with a Rising, Stable, Declining or Held chip, the current share, and the number of citations behind it. Held means we have the counts but not yet enough weeks of history to claim a direction.

Measured, never manufactured

This is cross-tenant on purpose: whether a source is losing ground with the engines is a fact about the open web, not about one customer, and computing it from a single account would make the number move whenever that account changed its prompts. No customer identifier or prompt text is read, only domain, week and a count. We report SHARE rather than raw counts, because a source can hold its count steady while the corpus grows and still be losing ground. Below eight weeks of stored answers we show the counts and withhold the arrow: a trend across three weeks is a line between two points.

Your on-site readiness score next to whether AI engines actually cited you, with the sample size behind both.