Sonar

Sonar is Perplexity’s retrieval and answer technology brand. It reflects the search-and-answer behavior that determines which sources are used and how the answer is shaped in Perplexity.

The important idea is retrieval quality. If the page is easy to fetch, easy to read, and easy to verify, the answer system has better material to work with.

For example, Ajey may want an AwesomeShoes Co. support page to be used in a Sonar answer about shipping. A clear policy page with direct wording is easier to ground than a general brand page. If the policy page gives one clean answer, Sonar has less reason to drift.

What helps

• Clear policy text.
• Easy retrieval.
• Short, direct passages.
• Pages that answer one thing well.

What hurts

• Broad pages.
• Vague wording.
• Hidden facts.
• Pages that need extra interpretation.

For AEO

Optimize the page for retrieval quality, not just keyword inclusion. A useful source is one the system can actually read and trust, similar to optimize for Perplexity.

Sonar-oriented optimization workflow

1. Identify query intents where source precision matters most.
2. Ensure target pages answer one clear question early.
3. Add supporting details and qualifiers near key claims.
4. Re-test source fidelity in Sonar-style summaries.

This keeps optimization tied to retrieval and answer fidelity.

Common failure patterns

• Policy pages with mixed promotional and operational copy.
• Missing date/scope markers for changing information.
• Overlapping pages competing for one intent.
• Evidence separated too far from recommendation text.

Quality checks

• Are selected passages faithful to original meaning?
• Do summaries preserve critical constraints?
• Is source attribution improving on fixed prompts?
• Are high-impact intents covered by dedicated pages?

Sonar performance improves when pages are engineered as precise, reusable sources with clear how Perplexity cites sources feedback loops.

Implementation discussion: Ajey (SEO lead), Priya (support lead), and the policy editor isolate shipping rules into answer-first sections, add scope qualifiers next to delivery-time claims, and run weekly Sonar-style prompt checks. They mark success when summaries keep policy constraints and citations consistently reference the intended sections.