LiftSignals is live. Every plan starts with a 30-day free trial — no credit card required. Start free →

How it works — every step between your store and a proven dollar

One line runs from your order history
to proven revenue.

No black boxes, no setup project, no analyst. Connect in minutes — eight platforms or a spreadsheet — and the line does the rest: one clean data model, 14 signal models scoring nightly, one-click audiences, governed sends, and a control group on every push. Walk it station by station.

First predictions on day zero · first proven number ~3 weeks later
1
Day 0 · 9:04 AM

Connect anything.
A spreadsheet counts.

One-click OAuth for Shopify, WooCommerce, BigCommerce, Square, Stripe, Lightspeed, Amazon, and Etsy. Not on the list? Upload a CSV — an interactive mapper matches your columns with a live preview.

Syncs are boring on purpose: re-running is always safe, a crashed sync resumes instead of duplicating, and a nightly refresh keeps everything current without you touching it.

8 platforms · CSV/Excel mapper · re-sync never duplicates
LIFTSIGNALS ShopifyOAUTH · SYNCED ✓ SquarePOS + ONLINE ✓ AmazonSP-API · THROTTLE-SAFE ✓ orders.csvMAPPED · 12,847 ROWS ✓
+ WooCommerce, BigCommerce, Stripe, Lightspeed, EtsyAd spend syncs too — Meta, Google, Amazon
2
Day 0 · 9:12 AM

Everything lands in
one clean model.

Whatever the source, your business normalizes into the same ten tables — customers, products, orders, line items, shipments, returns, acquisition, attribution. That's why a Square coffee shop and a Shopify apparel brand get the identical feature set with zero configuration.

Money is stored in integer cents with true margin after shipping and refunds — so every number downstream can talk profit, not just revenue.

10 tables · integer cents · true margin, not just revenue
customers products orders net_revenue_cents 412000 true_margin_cents 168400 order_items shipments returns acquisition attribution
Hover the cabinet — the orders drawer holds true marginRaw data never touched by re-scoring
3
Every night · 02:00

The night shift
scores everyone.

While you sleep, all 14 signal models retrain on your store alone — churn risk, purchase propensity, lifetime value, next likely product, reactivation odds, discount behavior, and more. Never pooled with other merchants; accuracy always on display.

Tonight's scores are diffed against last night's, so your morning starts with what changed: who crossed the churn line, who entered a buying window, who became a Champion.

14 models · your store only · overnight diff, every morning
14 MODELS YOUR STORE ONLY Maya K. Jordan T. Ana S. Maya K.churn 91↑ Jordan T.CLV $4.1K Ana S.buys 7d 12,847 IN 12,847 SCORED · 214 CHANGED
Every customer, every model, every nightAccuracy shown, not hidden — AUC, MAPE, drivers
4
Day 1 · 8:15 AM

Every number
is a button.

Any count, anywhere in the product, becomes an audience in one click — and the count you clicked is guaranteed to equal the audience you get. Cross two signals in the insight matrix and the sharpest segments appear: high churn and high value is the save-first list.

Audiences are living things: they re-resolve against tonight's scores, so "At-risk VIPs" is always today's at-risk VIPs — never a stale export.

Count = audience, guaranteed · re-resolves nightly · clone & tweak
CHURN RISK → ↓ LIFETIME VALUE 1,204 418 96 2,880 733 23HIGH × HIGH 4,102 1,566 310 Create audience (23) — exactly 23
Cross any two signals — the sharpest lists live in the cornersWhat you click is what you get
5
Day 1 · 8:16 AM

The gatekeeper
protects your list.

Before anything sends, every customer passes the gate: frequency caps, channel cooldowns, audience priority — and the model's no-contact verdict. The preview shows exactly who's suppressed and why, per customer, and what you preview is exactly what sends.

Then the push goes out to 11 live destinations — Klaviyo, Meta, Google, TikTok and more — with scores riding along, and a slice held back automatically for the part that comes next.

Caps · cooldowns · priority · no-contact is a real verdict
AT-RISK VIPS · 2,840 THE GATE CAPS · COOLDOWNS · PRIORITY · NO-CONTACT 2,528 SENTKLAVIYO · META · +9 MORE 4 suppressed emailed 4 days ago 312 HELD OUT CONTROL GROUP · FROZEN
Every send, governed — overrides exist for the flash salePreview = committed send, audited
While the line runs

Built to be trusted
when no one's watching.

Most of this machine runs at 2 AM with nobody in the room. It's engineered for that.

i.
Your data never mingles

Models train per-store, never pooled. Every read and write is isolated at the database level — enforced in every query, not just the interface.

ii.
Honest empty states

Not enough history to know something? It says so. No invented movement, no fabricated benchmarks, no "verified" noise below the statistical floor.

iii.
Failures are loud

A broken sync, a rejected export, an expiring connection — each surfaces as a notification with a fix path. Never a quietly stale dashboard.

iv.
The math can't be gamed

Holdouts are deterministic, windows freeze at send time, and preview-equals-send is checked by automated audits on every change to the engine.

Station 6 · The verdict

Three weeks later, the line
delivers its receipt.

Day 0
First signal, same session

Connect (or upload) and models run while you watch the sync. Onboarding ends with your first churn risks and a 30-day forecast.

Day 1
First audience live

The morning briefing lands, you push your first governed audience, and the control group is carved automatically.

Day ~21
First proven number

The window closes and treatment is compared to holdout. From here, every campaign compounds the proven total.

Proven
Lift report · push #241 · window closed
+$48,231
extra revenue that would not have arrived without the campaign — measured against the held-out control, not attributed.
Contacted (2,528)$96,410
Held out (312, scaled)$48,179
Evidence state✓ Proven
One click → management report · your logoGet your first receipt →
This is what the whole line exists to print
How long does setup actually take? +

Minutes, not a project. OAuth connects are one click; the CSV mapper previews your columns live. Models start running as soon as data lands — onboarding is designed to reach your first real prediction in a single session. Want to look around first? The live demo store is a full working workspace, no connection required.

What if my platform isn't on the list? +

A CSV of customers and orders is enough for the full experience — the interactive mapper matches arbitrary files onto the data model, imports are resumable, and a re-upload updates rather than duplicates. Custom connectors are an Enterprise conversation.

Why does the proof take ~3 weeks? +

Because it's a real experiment. The measurement window (typically 21–45 days by campaign type) has to stay open long enough for honest results, and it's frozen at send time so nothing can be gamed. While it runs, the panel shows "tracking" — never an invented number. When it says proven, it's proven.

Prove it

The line is already built. It just needs your store plugged in.

Connect in minutes — or upload a spreadsheet. First predictions today, first governed push tomorrow, first proven number three weeks out. Free for 30 days.

Start free trial See the live demo store →