How to retest off-site coverage: register, re-run the same prompts, compare the source pool
Off-site work is not finished when you publish: re-run the same prompts, check whether the cited source pool changed, and whether the new sources brought discoverability.
Key numbers
- Registration has quantified value: across 200 synthetic disruption episodes, registering offers in the same index lifted captured value from 46.8% to 98.6% — unregistered capacity caused requests to fail outright.
- Source structure varies by market: across 12 markets, 13 languages and 167,551 citations, about 18% of domains supply 80% of citations, and the top source was YouTube for Polish brands but a local business daily in Lithuania.
- Surfaces differ too: across four Chinese platforms and eight web/app interfaces, the same platform's app and web source sets differ systematically.
- Position has value: a randomised audit showed list position — a non-content factor — causally shifts recommendations, worth about $12 per night, so sources ranked higher keep collecting outsized recommendations.
Method and sample
Synthesis of verified studies: the open capacity pooling experiment (arXiv:2609.40296, value captured before and after registration), the cross-market source structure study (arXiv:2606.25787), the Chinese generative search audit (arXiv:2607.15771) and the hotel recommendation audit (arXiv:2606.16344, value of list position).
What it means for brands
Off-site results cannot be judged by feel: re-run the same prompt set, compare the cited source pool and your position within it, and see whether mention rates move with it. Discoverability comes from being registered in the right index and included in the right source set.
How SourceSight handles it
SourceSight pairs the off-site list with retesting: it names the platforms to cover per market, and after coverage the same prompts are re-run so the two snapshots can be compared on source composition and mention rate.
Limits
The evidence comes from different settings (supply chains, cross-market citations, hotel selection) and cannot be converted into one another; effect sizes depend on industry and market and need two snapshots from your own project.
References
Run it on your own project
What the same prompts return across 10 engines is only knowable by running them.
Related research
Same engine, different surface, different answer: test app, web and API separately
On the same platform and the same questions, app and web cite systematically different source sets — and the API shows something else again.
Methods · 2026-10-07How to design a retest you can trust: five moves so noise does not pass as progress
AI visibility is a distribution, not a point: fix the prompts, repeat the runs, add controls, paraphrase, validate by hand — or you cannot attribute the change.