How E-Commerce Teams Detect Revenue Drops and Pricing Errors Automatically
A pricing error on 40 SKUs costs you $12,000 per day. A mobile conversion drop costs you $8,000 per day. A traffic source dying costs you $5,000 per day. These things happen constantly in e-commerce — and they hide inside daily CSV exports that nobody has time to scan row by row.
By the time the weekly revenue review catches the problem, you've already lost 3–7 days of revenue. This guide shows how e-commerce teams use ThresholdIQ to catch these anomalies within hours instead of days.
The revenue leaks nobody catches in time
The pricing error that looks fine in aggregate
Someone updates prices in bulk — 200 SKUs across three categories. Forty of them get the wrong margin applied. Average selling price for those SKUs drops from $45 to $28. But across 2,000 active SKUs, the overall average selling price barely moves. The daily revenue report shows a 4% dip, which is within normal daily variance. Nobody investigates. Seven days later, someone notices the margin report is off by $84,000.
ThresholdIQ's Isolation Forest method would flag those 40 SKUs as structurally different from their category peers on day one. The deviation isn't based on a threshold — it's based on the SKU's price being statistically inconsistent with similar products.
The channel death spiral hidden in blended metrics
Google Ads CPC increases 35%. Conversion rate from paid search drops from 2.8% to 1.6%. But organic and email are performing normally, so the blended conversion rate is still 2.1%. The daily report says “conversion is slightly soft.” In reality, you're burning ad budget at 2x the normal cost while getting half the conversions. The ROAS collapse only becomes visible in the weekly channel-level breakout.
ThresholdIQ's correlation deviation method catches this: CPC and conversion rate for the same channel moving in opposite directions simultaneously is a signal, even when neither metric independently crosses a red line.
The mobile breakdown nobody sees
A checkout page update introduces a bug on iOS Safari. Mobile conversion drops from 3.1% to 0.8%. Desktop is unaffected. Since mobile is 55% of traffic, overall conversion drops from 2.7% to 1.9% — which looks like “a soft day.” The blended number masks a device-specific catastrophe. Three days later, engineering finds the bug. You've lost $24,000 in mobile revenue.
The pattern: E-commerce anomalies are almost always masked by aggregation. The total looks acceptable. The segment-level data screams. ThresholdIQ evaluates every column independently and flags segment-level deviations that get averaged out in summary rows.
What ThresholdIQ catches in sales data
- Revenue drops by product, category or channel — even when the total is within normal range
- Pricing anomalies across SKUs — products whose price is structurally inconsistent with their category
- Conversion rate shifts by device or traffic source — when one segment collapses but the blend looks normal
- Seasonal pattern violations — when a Monday doesn't look like a Monday, or a promotional period underperforms the seasonal expectation
- Correlated metric failures — ad spend up + conversion down + revenue flat = ROAS collapse, detected as a pattern even if no single metric crosses a threshold
How to set it up
- Export your sales data — from Shopify, WooCommerce, Amazon Seller Central, or any platform that gives you a CSV. Include as many dimensions as possible: product, category, channel, device, date.
- Upload to ThresholdIQ — drag and drop. Processing runs in your browser. No data leaves your machine.
- Review flagged anomalies — each one is graded Warning, Critical, or Emergency with the detection method that caught it.
- Set up email alerts — get notified when new anomalies appear in your regular exports.
7-day unlimited trial. No credit card. Works with any CSV export.