The headline impact, in six numbers
The six commercial KPIs every operator should know at a glance — revenue lift, conversion lift, CTR lifts at product and page level, demand coverage, activation rate.
Seven pages plus the Signal panel give operators the full window onto Flockr — what’s happening, what it means, and what to change. Skim the morning. Dig when it matters.
The platform map. Four covered below — Home, Analytics, Realtime, Settings. Three more plus the Signal panel live on their own platform pages.
At-a-glance commercial dashboard. The morning check-in — KPIs, catalogue pulse, activity trend, signal health.
↓Jump to HomeFour tabs of deep diagnostics — coverage, signals, messages, and per-product performance.
↓Jump to Analytics30-minute rolling window. Pipeline health, live shopper activity, top movers right now.
↓Jump to RealtimeDelivery mode, active surfaces, and read-only pipeline configuration. What operators control vs what the engine tunes.
↓Jump to SettingsPer-product demand intelligence — trending, fading, scarcity, ranks.
→Demand IntelligenceTemplate editor, surface variants, copy management — the only operator-editable page.
→Social Proof MessagingTwo-group session-level model, three lift measurements, empirical anchor.
→AttributionAI assistant, persistent panel across every portal page. Source-grounded, current to the hour.
→Signal AIAt-a-glance commercial dashboard. Four sub-sections, two clocks — 30-day historical for impact, 30-minute live for status. Read top-to-bottom for the full picture, or scan to the section that answers your question.
The six commercial KPIs every operator should know at a glance — revenue lift, conversion lift, CTR lifts at product and page level, demand coverage, activation rate.
Counts pulled straight from the demand intelligence layer. Refreshed every 30 minutes. Each counter links into its Demand tab for the full per-product picture.
Daily Flockr request volume against products messaged, plus signal family ranking — the quick read on what kind of evidence is feeding the engine.
The bottom-of-page strip that confirms sub-minute system health — visitors, requests, activation. If the Home page looks good and this strip is healthy, you can probably skip Realtime.
Historical view tells you how it’s going. Live view tells you it’s working. Two clocks, one page. Operators stay in the rhythm of the morning glance without losing the ability to confirm real-time health.
One surface. Top to bottom.
Coverage, signals, messages, products. Each tab is a different diagnostic question — and a different way to find what’s wrong before it hits revenue.
The diagnostic surface for activation. Read this when activation rate is lower than expected.
Signal usage and signal performance, side by side. Useful for understanding family mix and click rate by signal type.
Where messages get stuck. Selected but not rendered. Rendered but not visible. Visible but not clicked. Each transition is a different problem.
Which products convert with Flockr involvement, how shoppers progress through stages, and where Flockr is over-represented at conversion moments.
Diagnostics. Not just dashboards.
No date range. No history. Just a 30-minute rolling window onto live activity — sub-minute freshness, six views onto what’s happening this very moment. When something breaks, it shows up here first.
p50 / p95 on the Flockr request endpoint plus error rate. Persistent across every portal page — degradation is visible immediately, not just here.
Visitor count alongside the Flockr request volume those sessions are generating — the two numbers that should move together.
Products with the biggest selection-volume delta vs the prior window. Useful for spotting genuine demand surges as they happen.
Live distribution of signal families winning slots. The instant view onto what kind of evidence the engine is firing on right now.
Searches, product views, add-to-bags, purchases. What shoppers are doing — not just what Flockr is responding to.
Computed across the rolling window. Drift here often precedes a coverage problem you’d only see later in Analytics.
When something breaks, you see it here first.
The architectural split. Two operator-toggled controls — mode and surfaces. Four read-only configurations — surfaces, eligibility, thresholds, scoring — tuned by Flockr against observed conversion data.
Switch between rendering messages (live) and engine-running-silently (data). Effective on the next page load per visitor — no deployment needed.
Disable a surface and Flockr stops allocating slots there. Useful for phased launches and isolating diagnostic issues without touching delivery mode.
Operator toggles answer what should be running. Engine configuration answers how it should run. The substantive tuning is locked — deliberately read-only in the portal to prevent premature optimisation by operators without enough data to support it. Flockr tunes thresholds, eligibility, and scoring against observed conversion data; operators stay focused on the operational layer above.
Each active surface mapped to its journey stage — browsing, considering, or decided.
Signal families mapped to surfaces. attention not eligible on cart_drawer.
How much evidence a signal needs before a message can fire on its strength.
Relative weights used when ranking candidates competing for the same slot.
Toggle what runs. Trust what tunes.
Mode toggles, surface enablement, message copy — happen in the portal. Signal computation, scoring, message selection — happen in the underlying API on every page load.
The portal makes that engine activity legible and auditable. It does not run it.
Everything on this page — the dashboards, the diagnostics, the live windows, the read-only configs — is a view onto computation happening elsewhere. The work is on every request, at sub-second latency.
The engine itself — substrate, cohorts, selection logic, slot competition — lives on its own platform page. The portal is the surface; the engine is what makes the surface meaningful.
See the portal running on your store. Four pages covered here — Home, Analytics, Realtime, Settings — plus the three covered on other platform pages, plus the Signal panel on every screen.
Pipeline health, Signal pill, client selector — visible at all times. Degradation is one glance away regardless of which page you’re on.