How to rank in Perplexity answers
Perplexity has the smallest user base of the four major AI engines: roughly 45 million active users and 780 million monthly search queries by the end of 2025. But the audience is the one most brands want. Technical buyers, researchers, developers, and power users who ask sharp questions and expect cited answers. Per-visit conversion on Perplexity traffic is consistently higher than on ChatGPT or Gemini for B2B and technical DTC brands. This is the playbook we use at Soar to rank client content inside Perplexity answers for target brand and category queries in 2026.
Who uses Perplexity and why it matters
Perplexity is a real-time search engine wrapping a language model around a retrieval layer. Users ask a question, read the cited summary, and click through to one or two sources. That behavior is closer to Google than to chatbot traffic, which means brands cited inside Perplexity answers get real clicks instead of ambient mentions.
The audience is older, more technical, and more research-driven than the ChatGPT consumer base. We prioritize Perplexity for B2B clients whose buyers research before purchase. For mass-market consumer brands, Perplexity matters less. Audience fit is the first filter.
The three-layer retrieval pipeline
Perplexity's retrieval is more transparent than any other major engine. Ziptie's breakdown describes three layers that run sequentially when you ask a question.
Layer 1: initial retrieval via standard relevance scoring. Perplexity queries its index and pulls the broad candidate set using classical relevance signals. If Perplexity cannot crawl you, nothing that follows matters.
Layer 2: authority and relevance ranking. The candidate set gets reranked against authority signals Perplexity weights heavily: E-E-A-T factors, domain authority, content depth, and source diversity. Pages that pass layer 1 but fail layer 2 never make it into the answer.
Layer 3: XGBoost reranker for entity queries. For queries that mention specific entities (brands, products, people), Perplexity runs a machine-learning reranker that weighs the match between query entity and candidate source. "Best tool for X" and "compare Y vs Z" both trigger this layer, which is why it matters most for brand visibility.
The practical takeaway: ranking in Perplexity is not one optimization problem. You have to survive all three layers. Most DIY efforts focus on layer 1 content production and never address layer 2 authority or layer 3 entity matching.
The four credibility pillars Perplexity uses
Perplexity has publicly described its credibility scoring around four pillars (Authority Tech analysis).
Trustworthiness. Source reliability: domain age, HTTPS, accurate authorship, transparent editorial policies, absence of known misinformation patterns.
Authority. Weight in a specific topic area. A cybersecurity piece on a general-interest blog scores lower than the same piece on a dedicated security publication. Perplexity reads topical authority more narrowly than Google.
Corroboration. Whether a claim repeats across independent sources. A brand mentioned on one forum thread is weakly corroborated. The same claim on Reddit, in a press release, on LinkedIn, and in a GitHub README is strongly corroborated and more likely to be cited.
Provenance. Where the claim originated. Perplexity prefers primary sources over aggregators. The pipeline tries to trace restatements back to the original and cite that instead.
Corroboration is the pillar most brands underinvest in. You can write a great blog post, and Perplexity will treat it as single-source unless the information appears in multiple other places. Multi-surface content seeding is the single most effective Perplexity tactic we run.
The domains Perplexity manually boosts
Perplexity has been public about a handful of domains it boosts: GitHub, Amazon, LinkedIn, and Reddit (Data Studios analysis). These boosts apply across most entity queries. Get your brand discussed on those four domains and you feed Perplexity's retrieval at its strongest points.
GitHub. Open-source projects, SDKs, code samples, and READMEs with clear brand naming. If you sell a developer tool, this is non-negotiable. If you do not, consider publishing supporting assets (templates, example configs, integration tests).
Amazon. Product listings, review threads, and Amazon-hosted documentation. Critical for physical products. For SaaS and B2B, AWS Marketplace listings are the analog.
LinkedIn. Company pages, employee posts, long-form articles, and reshared press coverage. Under-used by most brands because they treat it as paid media instead of a citation surface.
Reddit. Category subreddits, AMAs, product comparison threads where your brand shows up by name. Reddit is the most important Perplexity citation surface we work on for almost every client. The full argument is in how Reddit became the biggest single source of LLM citations.
The stealth crawler controversy and your robots.txt strategy
On August 4, 2025, Cloudflare published an investigation showing Perplexity was using stealth crawlers to bypass robots.txt rules on sites that had explicitly blocked PerplexityBot and Perplexity-User. The Cloudflare post documents rotating user agents, IPs, and ASNs used to evade no-crawl directives.
If you block Perplexity at the declared-bot level, your content may still be crawled through undeclared agents. Our recommendation for most clients: allow Perplexity to crawl. The visibility upside outweighs the training-data concerns. If you run a publisher with serious IP protection needs, block the declared agents and monitor your logs for the patterns Cloudflare documented. The full walkthrough is in the AI bots robots.txt guide.
The Perplexity Publishers Program
On August 25, 2025, Perplexity announced the Comet Plus publisher revenue-share program: 80 percent to publishers, 20 percent to Perplexity, out of a $42.5 million initial pool (Perplexity announcement).
The program shifts incentives. Publishers who opt in get paid based on citations, which gives Perplexity a financial reason to surface their content more frequently. Brands that benefit most are the ones that can get onto partner publisher sites through earned coverage, guest contribution, or syndication.
This is also a hedge against the Reddit-Perplexity lawsuit filed in October 2025, which mirrors the Reddit-Anthropic suit from June. Expect the partner list to grow and the citation mix to shift toward program partners over the next 12 months.
How to measure Perplexity visibility
Perplexity is tracked natively by Parse, Profound, and HubSpot AEO Grader. Semrush AI Toolkit and Otterly.AI also include it. Build a prompt set of 50 to 100 queries covering brand, category, and comparison terms. Run them weekly and record mention rate, citation rate, source mix, and entity match (does Perplexity's summary correctly describe your product).
Entity match catches problems earliest. The layer 3 reranker is good at matching entities to queries, but it can confidently surface wrong descriptions if your content is fragmented across conflicting sources. Consolidate your brand messaging across every surface before you try to optimize for ranking. For audit methodology, start with how to audit whether your site is crawlable by AI bots.
Conclusion
Perplexity is the engine where the highest-intent audience lives and where the source-side mechanics are the most transparent. The three-layer retrieval, the four credibility pillars, and the manually boosted domains are all public information. What separates the brands that rank in Perplexity from the ones that do not is discipline: multi-surface content seeding, corroboration work across GitHub, Amazon, LinkedIn, and Reddit, and weekly measurement against a defined prompt set. The work is slow, but the audience is the one most brands would trade two competitors for.
How Soar saves you time and money
Brands that try to run a Perplexity-specific program internally typically give up around day 60. The reason is not that the work is hard. It is that the work is slow and almost impossible to measure without a dedicated tool. Corroboration-building across GitHub, Reddit, LinkedIn, and Amazon does not move the metric in week one. It moves the metric in week eight, and by then most internal teams have been pulled off to chase a more visible ChatGPT opportunity. We specialize in the corroboration work that actually moves Perplexity ranking, and we run the measurement cadence (through Parse) that proves when the interventions land.
The time savings are concrete. Most brands that come to us have already burned 40 to 60 hours on Perplexity content that never ranked, because they optimized for layer 1 and skipped layer 2 and layer 3. We start with the audit, identify where the layer-2 and layer-3 gaps actually are, and ship interventions that target those layers specifically. Within 90 days our clients are seeing measurable citation rate improvement instead of the flat lines most DIY efforts produce. Request a proposal and we will run your top 20 Perplexity prompts through Parse, identify the gaps at each layer, and propose a 90-day plan.