Speed and rendering decide eligibility
Perplexity's answer pipeline is optimized for latency. Pages that respond slowly or require client-side hydration to reveal their main content are frequently skipped, not because they are judged poor but because they did not return usable text inside the fetch window. This is a purely technical filter applied before any quality assessment happens.
Verify by requesting your key pages with scripts disabled and a short timeout, then reading the raw response. If the main article text is absent, prioritize server rendering for those templates ahead of any content work - it is the difference between being ineligible and being a candidate.
Response time deserves the same attention. A one-second improvement on a slow template can change citation frequency noticeably, because the crawler simply gets a complete document more often.
Freshness signals that actually work
Perplexity favours recent sources on any topic where recency plausibly matters, which is most commercial and technical topics. But bumping a date without changing the content is a losing strategy: repeated recrawls that find identical text reduce the value of the date signal for that URL over time.
A better policy is substantive refresh on a schedule. Re-run the numbers, add the new option that launched, remove the thing that shut down, and note what changed in a short update line at the top. The update line itself is often what gets quoted, because it is the most current specific statement on the page.
For pages built around data, publish the collection date alongside the figure. 'Measured March 2026 across 120 domains' is quoted far more readily than the same number with no provenance, and provenance is the single easiest quality upgrade available to most content teams.