Capture & Engage · Campaigns
Blast-and-pray sending tells you a campaign "did fine." Flash sends on a deliverability-hardened Amazon SES stack, tracks every open and click idempotently, and calls a winner only when a real significance test says so.
Chi-squared significance · not just whichever bar is taller
The problem
Send to everyone, hope it lands. Deliverability and unsubscribe compliance are an afterthought until inboxes start filtering you.
"Variant A won" — by a margin that's pure noise. Without a significance test, you're shipping superstition.
The moments that matter — a first purchase, a lapsing member — pass while someone waits to queue the next manual send.
How it works
Flash compiles your email once, renders per recipient with tracking, and sends through Amazon SES with a hand-rolled SigV4 signer, configuration sets, and one-click unsubscribe headers — then measures what comes back, without double-counting.
Honest A/B
Variant A
open 54%
Variant B
open 49%
Event-driven automation
14 of 15 liveTriggers fire from real lifecycle events — a receipt approved, a tier earned, 30 days of silence — so retention runs itself instead of waiting on a manual blast.
Opens, clicks, bounces, and complaints are deduped at ingestion — a webhook replay never inflates your numbers.
List-Unsubscribe + one-click POST headers on every send meet Gmail and Yahoo bulk-sender requirements.
Sent → delivered → opened → clicked, per-variant breakdown, hourly timeline, and top links — proxy-open adjusted for Apple/Gmail.
Why it's different
Plenty of tools send email and show you a dashboard. The difference is whether the numbers can be trusted and whether the next send fires on its own.
Typical approach
Blast-and-pray senders
One send to everyone; success is a vanity open rate.
Flash, by design
Deliverability-hardened SES with idempotent tracking and a real engagement funnel.
Typical approach
Naive A/B winners
Declare a winner off whichever bar is taller — noise wins half the time.
Flash, by design
Chi-squared significance with p-value and Bonferroni — a winner only when it's real.
Typical approach
Manual campaign calendars
Someone has to remember to send the lifecycle message.
Flash, by design
14 of 15 lifecycle triggers fire automatically from real member events.
AI & innovation
Flash's Campaign Decision LLM is fed only real engagement data — your top-clicked links, an honest peak-window timeline, A/B variants with their p-value and confidence, and per-pool coupon redemption — under explicit anti-fabrication rules.
The advisor reasons over your actual signals and returns a verdict, diagnosis, and levers — never invented lift figures.
A shared retry path and circuit breaker keep the advisor responsive when a provider wobbles.
Beyond the synchronous path, a distributed BullMQ + S3 chunk-render pipeline is built for high-volume batches.
Straight talk on scale: the wired send path runs synchronously up to 1,000 recipients per campaign. Larger batches route through the distributed chunked-render pipeline — a scalable architecture we'll size with you in a real environment, rather than quoting a headline number here.
Campaigns you can ship fast, measure honestly, and let run on autopilot for the moments that matter.
SigV4 SES
deliverability-hardened sending stack
p < 0.05
chi-squared A/B significance, not noise
Idempotent
open/click/bounce tracking never double-counts
14 of 15
lifecycle automation triggers live
See a campaign go out on the SES stack, a significance test call a winner, and lifecycle automation fire on its own.
