Skip to main content
Card class: Standard • Category: Revenue & Sales

​
At a glance

The automated watchdog on your sales line. It learns your normal daily rhythm over a trailing baseline, then flags any day that deviates far enough to be statistically unusual, in either direction. A sudden spike could be a viral moment, a competitor stockout, or a pricing error; a sudden drop could be a suppression, a stockout, lost Buy Box, or an ad budget that ran dry. The card exists so a meaningful move never waits for someone to notice it by eye on the Revenue Over Time chart.
What it doesPlots the sales series and marks points that deviate from the learned baseline by more than a set tolerance, surfacing both unusual drops and unusual spikes.
BaselineA trailing 30D model of your normal daily sales, accounting for the recent level and its typical variation. The alert fires when a day moves beyond roughly two standard deviations from that baseline.
DirectionBoth ways. A spike matters (capitalise, or catch a pricing error before it drains margin); a drop matters more (find and fix the leak fast).
Why automatedEyeballing the trend works for big, obvious moves but misses the gradual or the overnight ones. Statistical detection catches what a quick glance does not, and never sleeps.
Common drop causesSuppression, stockout, Buy-Box loss, an ad campaign pausing, a price change, or a category-wide demand shift. The card flags the when; the sibling cards explain the why.
Common spike causesA promotion, a competitor going out of stock, external press or virality, a coupon misconfiguration, or a pricing error that makes you the cheapest offer by mistake.
Relationship to the trendRevenue Over Time is the human-readable shape; this is the machine that watches it for you.
Time window30D baseline
Alert trigger>2σ from 30D baseline, a day beyond about two standard deviations of normal
Rolesowner, finance

​
Calculation

Calculated automatically from your Amazon Seller Central data. See the At a glance summary above for what the metric tracks and the worked example below for a typical reading.

​
Worked example

A pet-supplies seller. The 30D baseline is roughly 4,000/daywithatypicaldailyswing(onestandarddeviation)ofabout4,000/day with a typical daily swing (one standard deviation) of about 500. Figures are illustrative.
DateActual salesDistance from baselineFlagged?Likely cause
10 Apr 26$4,300+0.6σNonormal good day
11 Apr 26$3,700-0.6σNonormal soft day
12 Apr 26$6,800+5.6σYes (spike)competitor stocked out on hero ASIN
18 Apr 26$1,900-4.2σYes (drop)hero ASIN suppressed overnight
Baseline ≈ $4,000/day, σ ≈ $500, alert threshold ≈ ±2σ (±$1,000)
12 Apr spike  =  +$2,800  =  +5.6σ   →  flagged; investigate to sustain it
18 Apr drop   =  -$2,100  =  -4.2σ   →  flagged; investigate to fix it fast
Four things to notice:
  1. Both flags need action, for opposite reasons. The 18 Apr drop is the obvious one, find and fix the leak. But the 12 Apr spike matters too: a competitor stockout is a chance to grab rank and reviews while they are down, and it ends when they restock. Knowing the cause lets you press the advantage deliberately.
  2. The drop traced to a suppression. The 18 Apr fall lined up with a hero ASIN being suppressed overnight. The anomaly card caught it the same day; without it the seller might have lost a week to “sales feel a bit slow lately”. Cross-check New Suppressions (24h).
  3. It separates signal from noise. The 10 and 11 Apr wobbles (±0.6σ) are normal variation and correctly not flagged. The card only fires on moves big enough to be unlikely by chance, so the alerts stay credible and worth opening.
  4. A spike can be a costly mistake, not good news. Not every spike is a competitor stockout. A coupon misconfiguration or a pricing error that makes you accidentally cheapest also spikes volume, while quietly destroying margin. Always confirm the cause before celebrating a spike.

​
Sibling cards merchants should reference together

The anomaly flags the when. These explain the why:
CardWhy pair it with Sales Volume Anomalies
Revenue Over TimeThe trend this card watches. The anomaly markers sit on this same series.
New Suppressions (24h)A leading cause of sudden drops. Check it first when an anomaly is negative.
ASINs Stocking Out <7 DaysStockouts cause both drops (you ran out) and spikes (a competitor ran out).
Buy-Box Loss BurstA sudden Buy-Box loss produces a same-day sales drop the anomaly detector flags.
Revenue at Risk (live)If a drop traces to suppression or Buy-Box loss, this card quantifies the recoverable upside.
OrdersConfirms whether an anomaly is a volume move (orders changed) or a price/mix move (AOV changed).

​
Reconciling against Amazon Seller Central

Where to look in Seller Central: There is no Amazon-native anomaly-detection tile; this is a Vortex IQ analytic layered on the sales series. To verify any flagged day:
Reports → Business Reports → Sales and Traffic (by date) to confirm the actual sales on the flagged day, and the Sales Dashboard for a quick visual comparison of that day against its neighbours.
Amazon shows you the raw daily numbers; the value this card adds is deciding, statistically, which of those daily numbers is unusual enough to warrant attention, so you do not have to scan the series by eye every morning. Timing, settlement, and reporting-lag table:
TopicDetail
Partial-day false flagsThe current day is incomplete and will read low until it closes. The detector accounts for this, but the freshest point should be read with care to avoid a false drop flag.
Baseline warm-upThe model needs enough history to learn a stable baseline. For a brand-new listing or account, early flags are noisier until the 30D baseline matures.
SeasonalityA genuine seasonal shift (Prime-style event, holiday) can read as an anomaly because it deviates from the recent baseline. That is correct behaviour; it is unusual relative to normal, even if expected.
TimezoneDay boundaries use UTC; Amazon reports use marketplace-local time, so a flagged day may map to a slightly different calendar day in Seller Central.
Why our number may legitimately differ from Seller Central:
ReasonDirectionWhy
No native equivalentNot directly comparableAmazon has no anomaly tile. You reconcile the underlying daily sales, not the flag itself.
Baseline definitionJudgement callWhat counts as “anomalous” depends on the baseline window and threshold. A different window would flag different days.
Partial latest dayPossible false dropAn incomplete current day can look anomalously low until it closes.
Seasonal eventsExpected flagsKnown seasonal peaks may flag as anomalies because they exceed the recent normal; this is intended.
Cross-connector reconciliation against other connectors the same seller may run:
CardExpected relationshipWhat causes legitimate divergence
amazon.revenue_over_timeSame series, added detection. Every anomaly marker corresponds to a point on the revenue trend.None in the data; the difference is interpretation (which points are unusual).
shopify sales-anomaly cardsIndependent series. A DTC store’s anomalies are separate from Amazon’s. A shared promo can flag both.A channel-specific event (Amazon suppression, Shopify theme bug) flags only its own channel.

​
Known limitations / merchant FAQs

What makes a day count as an anomaly? A day that deviates from your learned baseline by more than the tolerance, roughly two standard deviations. In plain terms: a move big enough that it would rarely happen by chance given your normal daily swing. Small wobbles are ignored so the alerts stay meaningful. Why flag spikes, not just drops? Because spikes carry information and sometimes risk. A spike can be a competitor stockout to capitalise on, or it can be a coupon misconfiguration or pricing error draining your margin while volume looks great. Both deserve a look. Knowing the cause turns a spike into a deliberate decision. It flagged a day I knew was a sale, is that a false alarm? No, it is working correctly. A planned promotion genuinely deviates from your normal baseline, so it is statistically anomalous even though you expected it. You can mentally dismiss known events; the value is in the ones you did not expect. Why might the current day flag as a false drop? The current day is incomplete, orders are still landing. Until it closes it can read low. The detector compensates, but treat a same-day drop flag with a little caution and confirm once the day has matured. How long before it works on a new account? It needs enough history to learn a stable baseline (around the 30D window). For brand-new listings or accounts, early flags are noisier and become reliable as the baseline matures. The card flagged a drop, where do I start? Check the fixable causes in order: suppression (New Suppressions (24h)), stockout (ASINs Stocking Out <7 Days), and Buy-Box loss (Buy-Box Loss Burst). If it traces to one of those, Revenue at Risk (live) sizes the recoverable upside.

​
Tracked live in Vortex IQ Nerve Centre

Sales Volume Anomalies is one of hundreds of KPI pulses Vortex IQ tracks across Amazon Seller Central and 70+ other ecommerce connectors. Nerve Centre runs the detection layer; Vortex Mind investigates the cause when something moves; Ask Viq lets you interrogate any number in plain English. Start for free or book a demo to see this metric running on your own data.