Perplexity's Photon Engine: FAQPage Schema Doubles Citation Frequency

Perplexity published its Photon engine architecture in September 2026. The SEO community finally caught up in October — and the GEO implications are significant. FAQPage JSON-LD, which Google effectively killed for rich results in 2026, doubles citation snippet frequency on Perplexity. That's a 2× lift from one schema block.

What Perplexity's Photon Engine Actually Weights

Photon is Perplexity's internal retrieval engine — the system that decides which pages get pulled into responses before the LLM synthesizes an answer. Being good on Photon means getting into the candidate set at all. Miss it, and no amount of polished writing matters.

The five confirmed ranking signals, ranked by documentation clarity:

1. Query-term coverage— Photon's primary traversal signal. Pages that contain the query terms naturally distributed across headings and body copy rank higher than pages that mention them once. Keyword-native headings aren't a legacy SEO habit — they're a Photon signal.
2. Content freshness via dateModified timestamps — Recency-weighted with a ~90-day half-life. A page updated yesterday outranks an equivalent page last updated four months ago. The timestamp Perplexity reads is the dateModifiedfield in your Article JSON-LD — not your CMS's internal timestamp or a meta tag. If your JSON-LD dateModified doesn't update when you refresh content, Photon doesn't know you refreshed it.
3. Structural quality signals— Tables, FAQ blocks, and schema markup. This is where FAQPage JSON-LD enters the picture. A/B test data (Growth Marshal, 2025) shows FAQPage schema doubled citation snippet frequency — Perplexity pulls structured Q&A pairs directly into responses more than twice as often when FAQPage JSON-LD is present. Google's 2026 deprecation of FAQPage for rich results does not affect Perplexity behavior.
4. Domain-level authority signals— Present in Photon's ranking but less explicitly documented than the above. Consistent with other platform citation patterns: established entities outperform anonymous sites independent of content quality. Brand authority compounds here.
5. Answer-first structure in first 200 words — Roughly 20% of citation placement comes from the opening passage. Pages that bury the lead below a context dump lose that 20%. The answer to the likely query needs to land in the first two paragraphs.

Why This Matters

1. FAQPage JSON-LD is the highest-ROI schema implementation for Perplexity. One JSON-LD block — maybe two hours of work — doubles your citation snippet frequency. Google deprecated it for rich results; most practitioners stopped implementing it. That means your competitors have removed it while it still doubles Perplexity citations. This is a gap worth exploiting now.
2. Content refresh cycles are now a Perplexity-specific ranking variable.The 90-day freshness half-life means a page you haven't touched in five months is scoring at a fraction of its potential on Perplexity — even if nothing factually changed. Updating and re-dating content is now a maintenance task, not just editorial good practice.
3. Photon confirms platform divergence between Google and Perplexity. Google AI Overviews cares about structured data too, but through different signals. Perplexity weights dateModified and FAQPage explicitly. Google weights entity disambiguation and inline link prominence. Treating them as the same platform leaves Perplexity-specific wins on the table.
4. The 200-word threshold is binary. Answer-first below 200 words = Photon registers the source as a direct-answer candidate. Context-heavy intro above 200 words before the answer = citation placement drops ~20%. Structure your openings accordingly.

What to Do

1. Add FAQPage JSON-LD to every page targeting informational queries.
   • Identify 3–5 questions the page answers. Write them as natural-language questions, not keyword-stuffed phrases.
   • Write concise answers (50–150 words each). The answer needs to be self-sufficient — Perplexity may lift it directly.
   • Don't add FAQPage schema to transactional pages (pricing, contact, booking). Google may penalize over-application. Perplexity doesn't benefit from FAQ on non-informational content.
   • Google deprecated FAQPage for rich results — it still validates as schema. You're not breaking anything by keeping or adding it.
2. Fix your Article JSON-LD dateModified timestamps.
   • Audit which pages have Article schema. Verify dateModified updates automatically when content changes in your CMS.
   • For static pages (Next.js, Gatsby, Hugo), hardcode the dateModified and treat it as a required update whenever you refresh content.
   • Pages last updated 90+ days ago are deprioritized on Perplexity. A quarterly content refresh sweep — updating stats, examples, and the dateModified field — directly moves the freshness signal.
3. Rewrite page intros to front-load the answer.
   • The answer to the primary query must appear within the first 200 words. Not a hint at it — the actual answer.
   • Remove scene-setting paragraphs that don't contain substantive claims. "In today's competitive landscape..." costs you Photon citation placement.
   • Test your own pages: read the first 200 words. Would a Perplexity user get a satisfying answer from that passage alone? If no, rewrite.
4. Keep keyword-native headings even if they feel repetitive.
   • Photon's primary traversal signal is query-term coverage. Headings that contain the exact query terms help Photon confirm topic relevance quickly.
   • This doesn't mean keyword stuffing — it means using precise, literal language in H2s rather than clever metaphors. "How to implement FAQPage schema" beats "The structured way to answer questions".

Chad's Take

The uncomfortable truth is that most practitioners stopped implementing FAQPage schema the moment Google deprecated it for rich results — and nobody told them Perplexity still uses it and the effect size is 2×. That's not a small edge. That's a structural citation gap between sites that kept the schema and sites that removed it. Perplexity now drives a meaningful share of referral traffic in B2B and professional services categories. The companies treating Google and Perplexity as the same optimization problem are leaving half their AI citation potential on the table. Pick one high-value informational page, add FAQPage JSON-LD today, update dateModified, front-load the answer. That's 2-3 hours of work for a 2× lift on the platform your audience increasingly uses for research queries.

Sources

• Perplexity AI — Photon: Perplexity's retrieval engine — ranking signals, query-term coverage, freshness weighting, and structural quality factors, September 24, 2026. perplexity.ai
• SERPrecon — Analysis of Perplexity Photon engine signals and GEO implications, October 15, 2026.
• Growth Marshal — FAQPage JSON-LD citation frequency A/B test: 2× citation snippet lift on Perplexity, 2025. Via Perplexity AI Magazine.