Source influence is the measurable effect a specific source or source type has on the answers an AI engine produces, including which brands are named and how they are characterised. It is a concept defined and measured by blimpp. It separates two things that citation counting blurs: appearing in an engine’s source list, and actually shaping what the engine says.
In one sentence
Source influence asks not “was this source cited” but “did this source change the answer”, which is the difference between decoration and power.
How source influence is measured
Influence is assessed through two complementary observations across repeated runs. First, association: how answer outcomes such as brands recommended differ between responses where a source or source type appears in the citation set and responses where it does not. Second, carry-through: how often a claim or recommendation present in a cited source is reproduced in the answer’s substance. Recent research formalises this distinction as citation selection versus citation absorption, and the gap between the two is exactly what this metric tracks. Sampling follows the repeated-run protocol; source categories follow the source-type classification.
Worked example
Across 100 responses to a category prompt set, a community forum appears in the citation set 30 times. In 24 of those 30 responses, the forum’s most-endorsed brand is among the answer’s recommendations, against 9 of the 70 responses where the forum is absent. That asymmetry is the footprint of an influential source: its presence changes what the engine advises. Illustrative example: blimpp benchmark data is added to this page as studies publish.
Why source influence matters
Brands allocate effort by citation counts, but citation counts reward the wrong targets when a rarely cited source shapes answers and a frequently cited one merely decorates them. Influence mapping identifies the small set of surfaces where a change in content plausibly changes recommendations, which is where intervention budgets belong. It is also the concept that explains why third-party work can outperform on-site work for commercial prompts: the influential sources for buying questions are frequently not the brand’s own pages.
What affects a source’s influence
Evidenced factors: whether the source’s content is actually absorbed into answers rather than merely selected, per the selection-versus-absorption research. Factors with practitioner evidence: the directness with which the source states verdicts and recommendations, its perceived independence, its recency, and the density of specific, extractable claims it carries. These are treated as working hypotheses until tested.
Related concepts
References
- Zhang, K. et al. (2026), From Citation Selection to Citation Absorption: arxiv.org
Author: Harpal Singh · Last reviewed: 7 August 2026