Dispatches on physical-space analytics, AI-search visibility, and product-market intelligence — written by the Moonlight team.
Signal scoring across the corpus surfaced 12 genuine breakouts this quarter — here's how the liveness probes and press-velocity models spotted them before the noise did.
Track the same keywords on Google, ChatGPT, Perplexity, Claude, and Gemini at once and the divergence is striking. A breakdown of what Caposeo's unified workspace reveals about modern search fragmentation.
Table dwell tells you more about service quality and covers-per-hour than any POS metric — yet almost no operator tracks it. How Janus hits +22% table-dwell accuracy by separating staff from genuine guest time.
Answer-engine optimization has distinct signals, different latency curves, and a citation model that doesn't map onto Google's link graph. Why existing SEO platforms miss the point — and what a health score of 87 actually measures.
Most products don't die with a press release. They die in the data — declining liveness probes, zero press velocity, no new reviews. Across 18 categories, here's what the dying curve looks like before founders admit it.
Crowd-flow modeling in transit is hard: people move fast, cameras have blind spots, authorized staff blend into the public count. How Track Recovery and Camera Blind-Spot Tuning surface real congestion signal — and what 42ms latency makes possible live.
The most interesting signal lives at the intersections: tools migrating verticals, infrastructure turning into SaaS, developer utilities acquiring consumer traction. A guide to reading cross-category movement in the corpus.
A health score of 87 means little if you don't know which signals drive it. A breakdown of the four answer-engine models Caposeo monitors at once — and why citation share diverges even when organic rank holds steady.
Confidence and coverage scoring is the unglamorous module that makes the rest trustworthy. Why every Janus metric ships with an uncertainty band — and how a low-coverage camera angle quietly gets down-weighted.
A product going from zero to forty articles in three days looks nothing like steady background press. How rate-of-change on coverage volume separates a real breakout from a one-day spike.
When customers ask ChatGPT or Perplexity instead of Google, citation share predicts traffic before analytics ever sees it. What we track, why share diverges by engine, and how to read the donut.
The moat is operational intelligence, not biometrics. How Janus reads zones, flow, and dwell at 42ms latency without storing a single face — and why that constraint makes the product, not limits it.
Anomaly detection flags products breaking from their own history — not from an arbitrary threshold. A look under the hood at how 12 current breakouts cleared the bar while thousands of launches didn't.
One number from many signals invites the obvious question: which ones matter? A transparent breakdown of how crawl coverage, Core Web Vitals, structured data, and AI citations roll up into a single index.
Raw dwell time is an analyst's metric. How Janus translates guest-only table dwell into the operational numbers a floor manager actually acts on — without adding a single dashboard to their morning.
The sharpest field notes on physical space, AI search, and the product market — in your inbox, no filler. Unsubscribe in one click.
