For E-Commerce · Retail · D2C Brands

Detect revenue drops and
pricing errors before revenue bleeds.

Upload your Shopify export, sales report, or conversion data. ThresholdIQ's 9 ML methods automatically detect revenue anomalies, conversion rate shifts, pricing errors, and ad spend mismatches across SKUs and channels — no dashboards, no formulas, no rules to configure.

7-day unlimited trial · No credit card · Data never leaves your browser

$48B
in e-commerce revenue lost annually to pricing errors and fraud
9
ML detection methods running simultaneously
<60s
from file upload to full anomaly report
0
dashboards or rules to configure

Revenue leaks hide in plain sight

A pricing error on one SKU. A traffic drop on one channel. A conversion collapse on mobile. Each one costs thousands per day — and they're buried in your daily export.

💰

Silent revenue drops

One product category's revenue drops 25% but gets masked by a promotion spike in another. The daily total looks normal. You don't notice for a week — by which time the damage is done.

🏷️

Pricing errors across SKUs

A bulk price update goes wrong — 40 SKUs are listed at cost instead of retail. Average order value drops 15% but nobody flags it until the weekly review and three days of margin are gone.

📱

Conversion rate anomalies

Mobile conversion drops from 3.2% to 1.1% overnight. A checkout bug? A slow page? Desktop looks fine, so the blended rate barely moves. You lose three days of mobile revenue before anyone notices.

📊

Ad spend ⁄ revenue mismatch

Ad spend scales up but revenue stays flat. ROAS collapses on one channel while another masks it. The total looks justified until the month-end reconciliation reveals the true picture.

From export to anomaly report in 60 seconds

No analytics platform to learn. No SQL queries. Export your daily or weekly data from Shopify, WooCommerce, or your ad platform and upload.

1

Export your sales or analytics data

Daily sales reports, SKU-level revenue, conversion data by channel, ad spend reports, or customer cohort exports. ThresholdIQ supports .xlsx, .csv, .json and .xml — works with any platform export.

2

Upload — data stays in your browser

Drag and drop the file. All 9 ML detection methods run entirely in your browser. Your sales data, customer records and revenue figures never touch any server. No compliance risk, no data sharing.

3

9 ML methods detect anomalies automatically

SARIMA handles weekend and promotional seasonality so Black Friday spikes don't fire false alerts. EWMA catches sudden revenue drops. Correlation deviation flags when conversion, traffic and AOV all shift together — a platform issue, not a product issue.

4

Review anomalies graded by severity

Every flagged item is graded Warning, Critical, or Emergency with a breakdown of which detection methods fired. Export as CSV to append to your daily report, or generate a PDF for your weekly trading review.

Works across every e-commerce data type

If your data has a date and numeric metrics, ThresholdIQ finds what's abnormal — whether it's a SKU revenue line, a channel conversion rate, or an ad spend figure.

💰

Daily Revenue & GMV Monitoring

Multi-channel, multi-category daily revenue exports. SARIMA seasonal baselines learn promotional calendar patterns so weekend spikes and sale periods don't fire false alerts. EWMA detects sudden mid-week revenue drops in specific categories before they average out in the daily total.

🎯

Conversion Rate Anomaly Detection

Conversion rates by device, channel, landing page, and product category. Isolation Forest catches globally unusual combinations — mobile conversion collapsed while desktop is fine, or a specific traffic source showing abnormal session-to-purchase patterns indicating a checkout bug.

🏷️

SKU-Level Pricing & Margin

Product-level pricing, margin, and revenue per SKU. Stuck-value detection catches when a SKU's price stops updating — a sign of feed sync failure. EWMA flags sudden price changes across multiple SKUs simultaneously, catching bulk pricing errors within hours of occurrence.

📻

Ad Spend & ROAS Analysis

Daily ad spend and return by channel (Google, Meta, TikTok, affiliate). Correlation deviation flags when spend increases but revenue stays flat — a ROAS collapse that blended total spend hides. Trend detection surfaces gradual ROAS erosion across a campaign before the budget cycle ends.

🛒

Cart & Checkout Funnel

Add-to-cart rate, checkout initiation, payment success, and order confirmation by device and session source. Multi-window Z-score escalates a checkout success rate drop from Warning to Emergency as it persists — distinguishing a transient glitch from a payment gateway failure requiring immediate action.

📦

Returns & Refund Rate Monitoring

Return rate by product, category, and return reason code. DBSCAN cluster noise identifies SKUs with abnormal return profiles that don't match any normal return pattern — catching quality issues and fulfilment errors before they generate review damage and repeat return costs.

What ThresholdIQ finds in a daily sales export

Six channels in a daily revenue report — the kind of multi-row file where a single bad channel disappears into the total.

Channel / CategoryRevenueSessionsConv. Rate %AOV ($)ThresholdIQ
Organic Search$14,8204,2002.8$126Normal
Paid Search$6,3403,1001.1$186⚠️ Warning — conversion drop (EWMA)
Email$9,2101,8004.6$111Normal
Mobile App$4202,9000.3$48🔴 Emergency — checkout failure (multi-window)
Affiliate$3,1809603.1$107Normal
Direct$1,8402,4000.8$96🟠 Critical — correlated session + conversion

Mobile App flagged Emergency immediately — conversion rate collapsed from a normal 3.1% to 0.3% with AOV also anomalous, indicating a checkout flow failure. Direct flagged Critical for a correlated session quality and conversion drop. Both were invisible in the blended daily revenue total of $35,810 which looked within normal range.

How each ML method applies to e-commerce data

All 9 methods run in parallel across every metric in your file. Here's what each one specifically catches in sales and revenue data.

Multi-Window Z-Score

Evaluates each channel and SKU metric against rolling baselines at 50, 100, 200 and 500 periods. A checkout failure rate that persists across multiple windows escalates from Warning to Emergency — distinguishing a transient bug from a sustained platform failure requiring immediate escalation.

Primary severity driver
📈
EWMA Spike Detection

Catches sudden revenue drops, overnight conversion crashes, and instantaneous AOV anomalies. Fast-reacting — fires within the same reporting period as the event, before it compounds across multiple days and before the weekly review catches it.

Same-day detection
🌀
SARIMA Seasonal Residuals

Models Black Friday peaks, weekly shopping day patterns, and monthly promotional cycles. A legitimate Black Friday revenue surge won't trigger alerts. A revenue drop during a promotional period — when uplift was expected — fires immediately as genuinely anomalous.

Promotional calendar aware
🌲
Isolation Forest

Detects globally unusual combinations across all channels simultaneously — e.g. organic traffic and revenue normal while paid and direct both show abnormal session and conversion patterns together, indicating a platform or tracking issue rather than a product problem.

Cross-channel outlier
🔗
Correlation Deviation

Monitors relationships between traffic, conversion, and revenue metrics. When normally-correlated metrics diverge — traffic flat but conversion and AOV both dropping together — a pricing or UX issue is flagged. The multi-metric pattern reveals what no single-column review catches.

UX & pricing signals
📍
DBSCAN Cluster Noise

Identifies SKUs, products, or channels with behaviour that doesn't match any normal performance cluster. Catches specific product quality issues surfacing in return rates, individual affiliate partners behaving unlike all others, and outlier traffic sources with anomalous session behaviour.

SKU & channel outlier
🌙
Seasonal Baseline

Maintains separate normal ranges per day-of-week and time-of-period. Weekend shopping surges and Tuesday morning lulls are learned patterns. Alerts only fire when a metric deviates from its own expected level at that specific time — not from a flat average.

Day-of-week aware
📉
Trend Detection

Identifies monotonic drift in category-level revenue, gradual ROAS erosion across a campaign, and slowly declining repeat purchase rates. Flags the trend at Warning level weeks before it becomes a campaign optimisation emergency or a churn rate problem.

Campaign erosion warning
🚫
Stuck & Zero Detection

Repeated identical revenue or conversion values across multiple reporting periods indicates a tracking pixel failure, analytics disconnection, or data export issue. A sudden zero on an active channel fires Emergency — catching platform disconnections within hours rather than days.

Tracking failure detection

Your sales data already knows
something is wrong.

Upload your export. Nine ML methods run in seconds. Revenue anomalies, pricing errors and conversion drops surface instantly — graded by severity — with no dashboards required.

Just enter your email — no password required

Works with Shopify, WooCommerce, and any CSV or analytics export