Skip to content
Module 09 · AI Recommendation · live

Per-customer next-best-action. Routed automatically.

Lifecycle-stage classifier + per-customer channel preference + send-time optimisation + Bayesian discount learner. Wired into the comms-pipeline so the runtime dispatches the right message, on the right channel, at the right hour, for the right person.

Channels
0
email · WA · SMS · ads · push
Models
0
stage · channel · time · disc.
Lifecycle stages
0
new · active · at-risk · lapsed · won
Retrain
weekly
daily · enterprise
The problem

Most recommendation systems pick a product.

The harder problem is picking the channel, the timing, and the discount tier per customer.

Same channel for everyone

Email-first by default.

Every cart-abandon goes to email. Half your customers prefer WhatsApp; a third never open email. The default is leaving money on the table.

Same time for everyone

9am Tuesday for everyone.

The marketing calendar picks send-time once. Your customers' habits are individual; the marketing time is collective.

Same discount for everyone

15% off because it always was.

Some customers respond to 5%, some require 25%. Flat-discount policy over-discounts the easy converts and under-discounts the hard ones.

How it works

Profile. Score. Route. Attribute.

Every customer event runs through 4 models in parallel, then dispatches to the highest-confidence channel.

01 · Profile

Lifecycle-stage classifier

5 stages: new · active · at-risk · lapsed · won-back. Computed from order recency + frequency + monetary value. Updated on every order/refund event.

02 · Score

4 models, parallel

Per-customer Bayesian channel preference. Send-time per channel from historical engagement. Bayesian discount learner posterior. NBA scoring per content type.

03 · Route

Dispatch via comms-pipeline

Email, WhatsApp, SMS, paid social audience upload, ad bid up-weight - runtime routes to the highest-confidence channel. Per-shop throttle prevents spam.

04 · Attribute

A/B variant + close loop

When recommendation drives recovery, attribution-engine closes the variant loop. The model's posterior updates next retrain cycle. The feedback never stops.

The pipeline

Five sources. Four models. Five channels.

The exact runtime topology. Hover any node to inspect.

RECOMMENDATION · PIPELINE profile · score · route · attribute
Refresh · on-event · 8.4k NBA/min · live
customer events orders · cart · sessions order history RFM · cohort · LTV engagement log open · click · respond ab-test variants previous run posteriors shopify catalog product affinity RECOMMENDATION ENGINE idempotent · audited · multi-tenant 01 · PROFILE 02 · SCORE 03 · ROUTE Lifecycle stage RFM-based classifier · new· active· at-risk· lapsed 5 stages 4 Bayesian models channel · time · discount · NBA · channel pref · per-cust· send-time · per-channel· discount learner· NBA score per type· weekly retrain 4 models · parallel Dispatch highest-conf channel · email· whatsapp· sms· paid audience 145 RBAC permissions · 365d audit retention · snapshot + undo on every action email comms-pipeline · throttled whatsapp cloud api · template gated sms carrier · per-region ad audience meta · tiktok · sync
5 sources · 3 stages · 4 models · 5 channels
How it is used

Three ways this earns its place.

Capability, not customer outcomes - we have none to report yet.

4
channels routed per customer

Per-customer channel + send-time

Recovery does not have to be email-only. The engine infers a channel preference and a send window per customer, then routes across email, WhatsApp, SMS and ads. Whether that beats your current single-channel flow is an A/B question on your own list - which is how we would rather you tested it than on someone else's number.

Posterior
per-segment discount floor

Discounting to the level that converts

A flat recovery discount over-pays every customer who would have converted for less. The Bayesian learner builds a per-segment posterior on the discount actually needed and proposes a floor per segment. How much margin that returns depends on how blunt your current flat rate is.

5+
channels routed

Cohort-aware NBA wired to comms-pipeline

Lifecycle-stage classifier (new · active · at-risk · lapsed · won-back) drives the action set. NBA model picks the message + channel + time. The runtime dispatches via email, WhatsApp, SMS, paid social audience upload, or ad bid up-weighting.

By tier

What's available where.

CapabilityStarterProAgencyEnterprise
Lifecycle-stage classifier
Per-customer channel preference-
Send-time optimisation--
Bayesian discount learner--
Cart-recovery NBA + dispatch-
A/B-test attribution closure--
Ad-audience auto-upload--
Model retrain cadence-wkwkdaily
30-minute call

See per-customer routing on your data.

We connect to your Shopify + comms tools, run the channel + send-time models on a sample of 1,000 customers, and walk through 10 specific routing decisions live.