Unusual Movement Callouts

The Needs attention part of the insights block calls out movements in your report that are out of line with your store’s normal pattern: refund spikes, sales drops, rising discounts and coupon abuse. Each callout is a coloured row with the figure, how far it is from usual, and a small trend chart.

Deciding what counts as unusual is plain arithmetic on your own server — not AI. The AI service never chooses what is flagged, so every number in a callout comes from your store’s data. Callouts are part of AI Insights, so they only appear in reports with Show insights summary turned on, while AI is on.

What is watched

Only totals the report itself shows are checked, so a callout never quotes a number the reader cannot find further down the email. The options below are in the report builder under Report Totals and Report Details.

Total	Flagged when it…	Report option it needs
Net revenue, Gross sales	drops	Net Sales / Gross Sales
Orders, Items sold	drops	Orders / Items Sold
Average order value	drops	AVG. Order Value
New customers	drops	New Customers
Refunds, Number of refunds	rises	Refunds / Refund number
Coupon discounts, Coupon uses	rise	Net Discount Amount / Discounted Orders
Discounts as a share of sales	rises	Net Discount Amount
A single coupon code	is used far more than before	Net Discount Amount, Discounted Orders or Sales By Coupons

Good news — a revenue jump, a fall in refunds — is not called out. A good period already shows in the headline, the verdict badge and the totals, and repeating it would bury the problems.

How “unusual” is decided

Each total for this period is compared with the same total over the six periods before it, each the same length as the report period — so a weekly report is compared with the six weeks before, a monthly report with the six months before. A move is only flagged when it passes every test:

Enough history. At least three earlier periods with data. A brand-new store sees no total callouts until then.
Far from normal. At least two standard deviations away from the average of those periods.
Large in percent. At least 20% away from the average. Where a total never moved in the earlier periods, the bar is 50%.
Big enough to matter. Money totals must reach 1% of this period’s gross sales; counts must reach 3, or 1% of this period’s orders if that is larger. This stops one refund in a quiet week being reported as “refunds up 100%”.

Callouts are ranked by how far out of line they are, and only the strongest of each kind is kept — a drop in net revenue and gross sales is one callout, not two; likewise orders and items, refund total and count, and the three discount checks. At most three callouts are shown.

The discount and coupon checks

Check	Flagged when	Example
Discounts as a share of sales	Coupon discounts take at least 5 percentage points more of gross sales than usual. A coupon total can look normal next to a good sales week and still be a leak, which this catches.	Coupons took 12% of gross sales, against a usual 5%.
A single coupon	A code used at least 5 times that has either doubled its uses on the previous period, or is new this period and already on 30% or more of orders. Only the most-used such code is shown. This check compares with the previous period only, so it works on a new store too.	Coupon SUMMER20: Used 48 times against 12 the period before, taking $310.00 off. Check that SUMMER20 is not being shared or reused.

Reading a callout

You see	It means
Amber row	A problem worth a look.
Red row with an UNUSUAL badge	A severe move: three or more standard deviations from normal, a doubling against a flat history, or (for discounts) twice the threshold.
35% on your usual $1,200.00	How far this period is from the average of the earlier periods.
Usually there is none.	The earlier periods had none of this at all — for example the first refunds in weeks.
Most of it on Tuesday, Sep 23.	For refund and discount callouts: one day holds at least half of the period’s total, the way a refund run or a coupon posted online usually shows up.
Highest in 7 weeks. / Lowest in 7 weeks.	This period beats every earlier period shown in the trend chart.
Trend bars on the right	The earlier periods, then this one in a stronger colour.

For developers: tuning the callouts

All thresholds can be changed with filters, for example in a small custom plugin or your theme’s functions.php.

Filter	Default	Controls
`asre_report_anomalies_limit`	3	Most callouts shown. 0 for no limit.
`asre_report_anomalies_baseline_periods`	6	Earlier periods compared against. Below 3 turns the total checks off.
`asre_report_anomalies_min_score`	2	Standard deviations a move must reach. Receives the metric key.
`asre_report_anomalies_min_change`	20	Percent a move must reach. Receives the metric key.
`asre_report_anomalies_floors`	1% of gross sales / 3 or 1% of orders	Smallest money and count move worth reporting.
`asre_report_anomalies_min_discount_points`	5	Percentage points the discount share must rise.
`asre_report_anomalies_coupon_min_uses`	5	Uses before a single coupon can be flagged.
`asre_report_anomalies_coupon_min_share`	30	Percent of orders a new coupon must reach.
`asre_report_anomalies_positive`	false	Return true to call out good news as well.
`asre_report_anomalies_include_hidden`	Follows the report	Return true to check totals the report does not show.
`asre_report_anomalies`	—	The final list of callouts, to change or remove any.

// Show up to five callouts instead of three.
add_filter( 'asre_report_anomalies_limit', function () {
	return 5;
} );