Covers the Jun 9–10 working session. What shipped, what's wired but blocked, and the calls that need a decision.
Summary: the broker "2.1%" was confirmed a measurement artifact, not a leak — and we fixed measurement at the source with an ad-block/VPN-proof visitor cookie. Two new credibility features shipped to production (social-proof toast, auto-citations). Homepage CTA pivoted to consumer/business.
Social-proof toast. Live activity on provider pages (real-time) and the /vpn/ hub (aggregate "N people chose X today"). Grounded in actual /out clicks — counts use distinct first-party visitors with a ≥3 floor, so it never shows "1 person" or fabricated numbers. Single mute control, once-per-session per provider (no looping).
Auto-citations + citation schema. Every content page now auto-generates Wikipedia-style [n] footnotes, a numbered Sources section, and citation JSON-LD — built from the external links authors already wrote. Zero manual per-link work, ~442 pages covered, WCAG 2.2-clean. Distinguishes editorial sources from affiliate links automatically.
Homepage CTA pivot. Hero now offers two buttons — VPN for Consumers (→ /vpn/) and VPN for Business (→ /vpn/best-vpn-for-business/). Dropped the "go straight to our #1 pick" link (zero conversions).
Conversion tracking — fixed at the source. The broker funnel's "2.1%" is a tracking undercount, not a leak: real volume ~270 leads/mo, all reaching GHL. The client-side pixels on the thank-you page eat ad-block (~24% of the cohort) plus redirect-bail losses, and the loss is biased (skews desktop/tech-savvy) — so paid bidding optimizes on skewed data.
/out worker — server-set, HttpOnly, survives ad-blockers, VPNs, and no-JS. Serves three jobs: honest dedup key, affiliate sub-ID, and the join key for offline conversions.gclid we already store, with real deal value sendable later.Provider review schema audited. Pages already emit correct editorial Review schema (4.7/5, proper scale) — rich-result eligible today. Decided against adding aggregateRating (would require a review count we don't honestly have) and offers/price (volatile, merchant-controlled prices we don't auto-refresh → Google manual-action + bait risk). Both parked.
Domain-broker concepts polished per Michael's feedback (4 concepts on concepts.vpn.com).
priceValidUntil). It's the prerequisite for ever putting offers in schema or price chips in the SERP. Decision: do we invest in keeping 22 providers' prices current?