How the numbers work
Markdown reports and historical ranges
The dashboard reads validated build-time projections of local intelligence reports. Originals stay local. Public editions contain recognized metrics and separately reviewed commentary, excluding operational workflows and private account links. Range links use each source's latest reported date as their historical anchor, not today. A 28-day aggregate is an available period, never an exact 30-day or lifetime total. Overlapping snapshots are never added; missing daily data is never reconstructed. Growth comparisons are withheld because the launch baseline is incomplete.
Sources and publication
PostHog measures recorded website activity and available field performance samples. Google Search Console measures Google web search visibility and clicks. Vercel supplies historical deployment status; no live availability is inferred. Data is published as reviewed aggregate snapshots; there is no automatic refresh schedule yet. Each available source shows its own reporting period, time zone, and last successful update. Missing or rejected snapshots show Unavailable, never zero.
Visitors, pageviews and sessions
Unique visitors are distinct PostHog visitor IDs for the complete selected period, not a sum of daily visitors. Pageviews count recorded pageview events. Sessions count distinct session IDs using the project's session settings. A person may use several devices or clear cookies, so visitors do not equal people. Ad blockers, consent settings, network failures and disabled JavaScript can prevent collection.
Organic search
Impressions count appearances in Google web search. Clicks count clicks from those results. CTR is total clicks divided by total impressions, not an average of daily percentages. Average position is impression-weighted and lower is better. Search Console may delay, revise or omit data for privacy. Its daily dates use Pacific time (America/Los_Angeles); other source time zones are stated on their snapshots. Search clicks must not be added to PostHog visitor counts.
Performance
LCP (milliseconds) measures largest content paint, INP (milliseconds) measures interaction responsiveness, and CLS measures layout instability without a unit. Published values use the 75th percentile of eligible real production samples for the stated period. They are not synthetic Lighthouse scores. Small samples and device mix affect comparisons. No uptime percentage is inferred from analytics.
Filtering and privacy
Only reviewed aggregates are public. No individual visitor IDs, raw URLs with query strings, contact details, revenue, client identities or event payloads are published. Search queries are withheld. Publishers must suppress sparse or identifying breakdowns, normalize landing paths, and record bot and internal-traffic exclusions with each report. No bot filtering is claimed until verified in the source configuration; unfiltered automated traffic can inflate counts.
Contact intent and comparisons
contact_whatsapp_click, contact_booking_click and contact_email_click record one explicit contact action with its placement. General automatic click capture may also exist in PostHog; use only these named events for contact-intent reporting. A click does not prove that a conversation opened, a booking completed, or a qualified lead arrived. Compare complete, equivalent periods from the same source; do not average percentiles or sum daily unique visitors. All time means the available published source history, not an assumed launch date.
Corrections and reports
A failed refresh leaves the last reviewed snapshot in place, with its original timestamp. Historical reports are manually reviewed and published only when they contain meaningful analysis. Reports state the period, source metrics, prior-period changes where comparable, website updates, observed trends, caveats and next steps. Correlation alone is not proof that a website change caused growth.