Pricing has two columns: regular_price and sale_low. A sale_frequency field ("often", "twice a year", "rare") informs the prose surfaced on the page. When Soundtoys runs its annual 75 percent off sale, the page already says "often around $99 during Black Friday" because that note is data. The corpus does not chase every sale; it tells buyers the realistic price band.

A plugin row carries a current_version field and an old_version field. When FabFilter ships Pro-Q 4, edit the rows, and the page automatically surfaces the new version as the recommended pick while noting the previous version's availability. Updates flow from one edit, not a corpus-wide rewrite.

Yes. Plugins are tagged with vibe descriptors. A page filtered on vibe=clean ranks differently than one filtered on vibe=vintage. Same plugin sheet, two leaderboards, two URLs, no duplicated data. The character-EQ page surfaces Pulsar Massive Passive; the surgical-EQ page surfaces FabFilter Pro-Q 4.

Bundles are their own rows with a contained_plugins array. The page can surface bundles when they represent the best value for the category, or filter to individual plugins for buyers shopping per-tool. Bundle pricing tracks separately, so a Black Friday bundle drop reflects accurately.

Pricing model is a field. Subscription plugins surface a monthly_rate alongside their headline price. The leaderboard can sort on either model, and the page can render a TCO comparison over a chosen ownership horizon driven by the user's intent.

Yes. Each plugin row carries a short_verdict field; each category row carries a long_form_intro that renders above the leaderboard. The structured ranking handles the discovery; the editorial prose handles the recommendation. Both live in the same sheet, both update without rewriting the page.

Free plugins (Valhalla Supermassive, TDR Nova, Vital) live as rows with regular_price=0. The leaderboard sorts them based on the same reputation score as paid plugins. A separate page group at /audio-plugin-for/free-{slug}/ filters to only free entries when a buyer specifically wants no-cost recommendations.

Tracked as fields: cpu_load (low/medium/high), supports_arm, supports_native_silicon, supports_vst3, supports_au, supports_aax. Pages can surface the compatibility matrix and filter rankings by what runs on Apple Silicon or what stays light on the CPU for tracking sessions.