Capture & Engage · Dashboard
Most dashboards report what happened and leave the "so what" to you. Flash reads your real-time scan-to-redemption and loyalty data, finds the weak step, and puts the next move on the same screen — over one verified member, not five disconnected tools.
QR Scans
+12% WoW
Captures
+8% WoW
Capture Rate
+3pt WoW
Redemption
-2% WoW
Coupon issued-to-redeemed is the funnel's weak step. Re-engage the 14-day no-redeem segment with a 48-hour offer.
The problem
Identity, capture, loyalty and channel metrics scattered across tools — so the full picture lives in nobody's head.
Charts go up and to the right, but the dashboard never says which lever to pull this week.
When a funnel step quietly leaks, you find out a quarter later — after the lift is already gone.
How it works
Every KPI carries a real week-over-week delta and a multi-window sparkline, computed from period-over-period aggregates — not mocked. Store-managers see only their assigned stores.
The headline view — what moved this week, at a glance.
QR scans, captures, capture rate and the scan-to-coupon funnel.
Active members, redemption, tier pyramid and 8-week retention.
Full-funnel conversion and channel mix across your stores.
Capture funnel · scan → redeemed
Per-step conversion with automated bottleneck diagnostics — the board names the weak step, it doesn't make you hunt for it.
RULE ENGINE
Detect the pattern
Deterministic rules scan the snapshot, then cooldown + compliance dedup decide what's worth surfacing.
LLM · GUARDED
Draft the next action
The LLM phrases the recommendation; guardrails check it — with a rule-based fallback if the model call fails.
A real retention curve from genuine SQL — enrolled-before-window vs. still-active, not a sampled estimate.
A hand-built SVG map with a custom lat/lng projection and bubble sizing per geocoded store.
Per-service caches with a 1-hour TTL; every scan and capture writes daily stats and invalidates the cache.
Deeper RFM, customer-lifetime-value and health-score intelligence lives at the individual-member level — powering the member detail view, not the four boards. We keep that distinction honest so the dashboard stays trustworthy.
Why it's different
Incumbents make you assemble identity, segmentation, analytics and decisioning across separate products. Flash unifies the member as one verified identity — then surfaces the next best action on the same screen instead of in a separate decisioning suite.
Typical approach
Reporting dashboards
Show trends and stop there — the decision is left to you.
Flash, by design
A rule-plus-LLM panel that recommends the next best action, on the same screen.
Typical approach
Separate decisioning suites
Bolt propensity scoring on as another product to log into.
Flash, by design
The action sits beside the funnel that triggered it — no context-switch, by design.
Typical approach
Fragmented analytics stacks
Count the same customer twice across disconnected tools.
Flash, by design
Every board reads over one verified member, unified across store, QR, Shopify and more.
AI & innovation
The market is moving from dashboards that report to systems that decide. Flash's AI Actions panel is a genuine rule-engine-plus-LLM pipeline — the agentic "Decide" surface of the retention loop.
Deterministic rules detect the pattern from a real data snapshot; the LLM drafts the recommendation in plain language.
Cooldown and compliance dedup stop nagging; guardrails vet every recommendation before it reaches you.
If the model call fails, the panel falls back to rule-based recommendations — it always returns a usable next step.
Less time reading charts, more time pulling the right lever — with the weak step already flagged.
4 boards
Summary, Engage, Loyalty, Omnichannel — tier-gated
Real WoW
deltas + sparklines computed from real aggregates
8 weeks
of genuine cohort retention from live SQL
Rule + LLM
next best action, with a rule-based fallback
We'll walk the four boards, the live store map, and the next-best-action panel — on a real account.
