RetainIQ predicts which customers are about to disappear, then automatically fires the right offer, at the right moment, through the right channel — before they ever decide to leave.
14-day free trial · no card · five minutes to connect your store
Recovered revenuerolling 12 months
$224,000Annual revenue recovered
34%Of at-risk customers saved
96.2%Model accuracy, held-out data
7M+Events scored monthlyOrders, sessions, tickets and refunds across every connected store.
$480Average lifetime value savedWhat a single recovered customer is worth, on average, across our accounts.
4 daysTo first useful predictionThe model needs a few days of your history before its scores are worth acting on.
5 minTo connect your storeOne-click install on Shopify, BigCommerce, WooCommerce and Magento.
The pipeline
Five stages. One engine.
Every customer moves through the same five stages, continuously. There is no campaign to build and no list to segment by hand.
01
Data Sync
Orders, sessions, tickets and refunds stream in from your store the moment they happen.
~90s per sync
02
Feature Build
Recency, cadence, basket drift and support tone become 140+ features per customer.
140+ features
03
Risk Scoring
The model returns a churn probability and a reason code for every active customer.
96.2% AUC
04
Offer Decision
Discount depth is capped by predicted lifetime value, so margin never pays for a save.
LTV-gated
05
Dispatch
The right message goes out on email, SMS or push, at the hour that customer opens.
4 channels
Scored this morning8,412 active
CUST-48201risk 0.9114 orders · last 62 days ago · 2 tickets open
CUST-19477risk 0.846 orders · last 38 days ago · basket down 40%
CUST-77310risk 0.5222 orders · last 19 days ago · cadence slipping
CUST-30885risk 0.173 orders · last 11 days ago · steady
CUST-61204risk 0.0931 orders · last 6 days ago · steady
The model
The model that knows before they leave
Most stores find out a customer has gone when the quarterly numbers come in. By then winning them back costs more than keeping them would have.
Your own order historyTrained on your store, not an industry benchmark. A 40-day gap means something different for coffee than for furniture, and the model learns which you are.
Frequency breakdownCadence, basket drift, session depth and support tone are weighted together, so a quiet month on its own does not trip an alarm.
LTV-capped offersPredicted lifetime value sets the ceiling on what the engine is allowed to spend winning somebody back.
What you get
Built for merchants who sell seriously
Six parts, one engine. Nothing here needs a data team to operate.
Precision scoring model
Gradient-boosted trees trained on your own order history, not an industry average. Every score ships with the three signals that drove it.
reason codes included
LTV-gated offer engine
Discount depth is capped by predicted lifetime value. A customer worth $90 never gets handed a 40% code to come back.
margin protected
Multi-channel dispatch
Email, SMS, push and on-site. The engine picks the channel each customer actually responds to, at their open hour.
email / sms / push / web
Universal platform support
Shopify, BigCommerce, WooCommerce and Magento out of the box. Anything else connects through the REST API.
4 native, 1 API
Revenue attribution
Every save is tied back to the offer that caused it, with a holdout group running so you can see the real lift.
holdout built in
AI analyst, built in
Ask why a cohort is slipping and get a written answer with the numbers behind it, not another dashboard to read.
powered by Claude
Data residency
Your data never leaves your stack
RetainIQ runs in your own cloud account, in the region you choose. Nothing is pooled across merchants and nothing trains a shared model.
AES-256 at rest
TLS 1.3 in transit
GDPR
SOC 2 Type II
PCI DSS
Your region, your keys
Merchant results
Results that compound
★★★★★
"We were losing about $9,000 a month to silent churn. RetainIQ caught 23 high-risk customers in week one. Three of them had spent $400+ each historically."
SPSana PriyanFounder, Thread Nine
★★★★★
"The LTV-gating is what sold me. We were training customers to wait for discounts before RetainIQ. Now our discount spend is 40% lower with better results."
DODmitri OakesHead of Growth, Northbay
★★★★★
"Set up on a Tuesday, first win-back campaign fired Thursday. By Sunday we had recovered $7,340 from 16 customers we'd have written off."
KRKaren RuizCOO, Elder & Finch
Pricing
Plans that pay for themselves in days
One recovered customer is worth $480 on average. Every plan below costs less than that in a year.
Spark
Spark
For stores finding their first thousand customers.
Running several storefronts, a marketplace, or more customers than Command covers? We will size the model, the infrastructure and the price around what you actually run.
All plans include the 14-day trial. No card up front, and no charge until the model has produced its first scores.
The annual case
Why the annual plan pays for itself twice
Once on the billing, and again because the model has a full year of your history to learn from instead of restarting every renewal.
Retain plan, one yearCost
Monthly billing, 12 × $30$360.00
Annual billing, paid once$268.20
You keep$91.80
That is before a single save. One recovered customer at the $480 average covers either plan for more than a year.
Cash out, by month
Retain plan
JFMAMJJASOND
Annual, paid onceMonthly, every month
Before you decide
Questions before you commit?
How accurate is the churn prediction?
The model holds 96.2% AUC on held-out data across our live accounts, measured against customers who did in fact stop ordering. Accuracy climbs for about six weeks as it learns your store specifically, and every score carries the three signals that produced it, so you can sanity-check it rather than take it on faith.
What is the difference between monthly and annual billing?
The features are identical. Annual is billed once at the start of the year and works out roughly 25% cheaper per month. Monthly is billed every month and you can stop whenever you like.
Will sending discounts train customers to wait for them?
That is exactly the failure mode the offer engine is built to avoid. Discount depth is capped by predicted lifetime value, offers only fire for customers the model flags as genuinely at risk, and a holdout group stays untouched so you can see what you would have earned anyway.
How long does setup take?
About five minutes to connect your store, then roughly four days before the first predictions are worth acting on, while the model builds history. Most accounts have their first recovered order inside the first week.
Can I cancel the annual plan?
Yes. Cancel in the first 30 days and the unused portion is refunded in full. After that the plan runs to the end of its term and then stops. Your exported data is yours to keep either way.
Where does my customer data live?
In your own cloud account, in the region you choose. RetainIQ reads what it needs through scoped credentials and writes scores back. Nothing is pooled across customers and nothing trains a shared model. AES-256 at rest, TLS 1.3 in transit.
Start today
Stop watching customers vanish.
Connect your store in five minutes. See your first at-risk list inside a week, and decide from there.
.png){kind=small}