Summary
| Recommendation |
|---|
| Source diversity measures whether AI answers draw from a balanced mix of owned pages, third-party profiles, review sites, support docs, community threads, publisher pages, and competitor sources. |
| It does not tell you which source is most influential, whether the answer is right, or whether a diverse source mix is actually desirable for that prompt. |
| Use it to identify overreliance. If a category prompt is mostly shaped by support docs, Reddit threads, or competitor pages, the source mix may be revealing a content gap. |
| Which source type is overrepresented or missing, and what does that say about the evidence AI systems can find? |
Metric details
| Criterion | Value |
|---|---|
| What it measures | Source diversity measures whether AI answers draw from a balanced mix of owned pages, third-party profiles, review sites, support docs, community threads, publisher pages, and competitor sources. |
| What it misses | It does not tell you which source is most influential, whether the answer is right, or whether a diverse source mix is actually desirable for that prompt. |
| How to use it | Use it to identify overreliance. If a category prompt is mostly shaped by support docs, Reddit threads, or competitor pages, the source mix may be revealing a content gap. |
| Bad interpretation | A bad interpretation is assuming more source types are always better. Some prompts should rely heavily on neutral third-party sources; others should rely on current owned explanations. |
| Next diagnostic question | Which source type is overrepresented or missing, and what does that say about the evidence AI systems can find? |
FAQ
How should teams use source diversity?
Use it to identify overreliance. If a category prompt is mostly shaped by support docs, Reddit threads, or competitor pages, the source mix may be revealing a content gap. For example, use source diversity to decide whether the next step is monitoring, source review, answer interpretation, or a specific content update. Read the metric beside the actual source material, because the number alone cannot tell whether a cited or recurring source is useful, stale, overrepresented, or misleading.
What does source diversity miss?
It does not tell you which source is most influential, whether the answer is right, or whether a diverse source mix is actually desirable for that prompt.
What is the next diagnostic question?
Which source type is overrepresented or missing, and what does that say about the evidence AI systems can find?
What decision should this metric inform?
Source Diversity should inform the next diagnostic step: Which source type is overrepresented or missing, and what does that say about the evidence AI systems can find? For source diversity, if the team cannot answer that, keep the signal in review instead of turning it into automatic content work.