Perplexity (Sonar)
By Şamil Quliyev · Updated 2026-09-15
Perplexity is an answer engine whose Sonar models retrieve before they write: search first, answer second, with numbered citations attached to the claims. Because grounding is the default rather than a decision, its answers are the closest thing in this category to a direct read on what the live web says about a brand.
How it picks sources
Sonar favours pages that answer the question explicitly and are easy to attribute — documentation, comparison pages, well-structured articles and recent posts. Aggregators and community threads surface often, which is why a brand can appear in Perplexity answers through a third-party page far more often than through its own site.
Grounding behaviour
Effectively always grounded, which removes the memory-answer ambiguity but introduces a different one: the retrieved set changes constantly, so two runs minutes apart can cite different sources while making the same claim. Sample size matters here for source stability rather than for grounding status.
How to track it
Call the Sonar API with the same prompt set you use on other engines, keep the citation array with each answer, and separate two numbers: how often the brand is named, and how often the brand's own domain is the cited source. The gap between them is usually the most actionable output of tracking this engine.
Measurement mistakes
- Reading a high mention rate here as site authority. Perplexity may be mentioning you because a review site said so, not because it read your page.
- Comparing its citation rate directly with engines that cite sparingly — the engines have different citation habits, so only the trend per engine is comparable.
- Tracking only one phrasing of a question. Sonar is sensitive to phrasing, and a near-synonym prompt can return a different source set.
What TAMGIO measures here
This engine is part of TAMGIO ongoing tracking: prompts are sampled on it every cycle.
Frequently asked questions
- Why does Perplexity cite competitors on our own brand queries?
- Because it retrieves what best supports the sentence it is writing. If a comparison article covers your category more explicitly than your own pages do, that article gets cited even on a query about you.
- Is Perplexity worth tracking separately from ChatGPT?
- Yes — the retrieval behaviour and citation habits differ enough that the two engines regularly disagree about the same brand, and one engine is not a proxy for the other.
