Signals & AI
Anomaly detection and prescriptive signals
A feed of anomaly alerts on every key metric, plus a prescriptive engine that combines several metrics against a rule to name a probable cause and a next step.
The problem
Most founders find a problem days late, by noticing the bank balance, then spend an afternoon working out which channel, SKU, or campaign moved.
How Sellrics does it
Sellrics runs anomaly detection against each metric’s expected range for the period (backend/api/anomalies.py) and a 10-rule prescriptive signal engine (backend/api/signals.py) that fires when a combination lines up — for example margin down + return rate up + same SKU. Root-cause drilldown (backend/api/root_cause.py) then decomposes the change into the channels, SKUs, and campaigns that caused it.
What the numbers mean
An anomaly alert means a metric has moved outside its expected band for this period. A prescriptive signal is a named pattern with a suggested action, not just a number. Root cause attributes a change proportionally to its contributors.
A worked example
Illustrative example- Scenario
- Contribution margin drops 4.1 points week over week.
- What Sellrics surfaces
- The alert feed shows the drop; the prescriptive signal ties it to one SKU whose return rate jumped from 18% to 31%; root cause says that SKU accounts for 2.7 of the 4.1 points. Two more signals flag a Google Ads CPA up 22% and three campaigns over target.
- The move
- Pull the SKU’s listing photos and reviews that day, pause the over-target campaigns, and check the supplier batch — instead of starting an investigation from zero.
Insight Summary
Sample · 8 alertsHigh priority
2
Warnings
3
Opportunities
2
Positive
1
Sample data — the alert summary and feed you get each morning.
See it with sample data
Fictional sample dataA view from the Sellrics dashboard, on fictional sample data — no store connected, nothing stored.
Needs Attention
Sample data · live alert examples8 SKUs missing COGS data
4 SKUs below 7 days of stock
1 SKU has negative net margin
3 SKUs missing listings on a channel
Top 3 products = 71% of revenue
Amazon buy box slipping on SE01
Contribution margin up 2.1 pts
Amazon conversion up 0.8 pts
Insight Summary
Sample data · 8 curated alerts across four severities
2
2
2
2
All Insights
Sample data · no real data loaded8 SKUs missing COGS data
≈$3.4MMap COGS for the affected SKUs to unlock true margin on $3.4M of annual revenue.
4 SKUs below 7 days of stock
≈$96KSE10, SH03, SH07 and FBA-SE01 have under a week of inventory — reorder before stockout.
1 SKU has negative net margin
≈$25KFBA-SE01 (Loss Leader) is selling below cost — review pricing, COGS, or ad spend.
3 SKUs missing listings on a channel
SH03, AM05 and FBA-SE01 have no active listing on one expected channel — review the Listing Gaps table on the Operations tab.
Top 3 products = 71% of revenue
Revenue concentration risk — consider diversifying the catalog and bundles.
Amazon buy box slipping on SE01
≈$30KPremium Widget buy box is at 91% — small price drops can recover lost sales.
Contribution margin up 2.1 pts
Fee-aware margin improved vs prior month — pricing and fee changes are paying off.
Amazon conversion up 0.8 pts
Conversion rose to 4.0% on updated listings — keep the new titles live.
Detail actions are disabled in this demo. Connect a store to open the real insight panels.