Conversion & Checkout Funnel Radar · Brief 02
Black Friday · paid conversion collapsed while impressions held
Hardlines · Midwest · 2025-11-28 · impressions flat → orders down 45% → demand diverted
The one line you configure
Paid conversion on the Hardlines campaigns fell hard this morning while impression delivery stayed on plan. Is that the creative, the audience, or something the ads are pointing at?
In a nutshell
Something the ads are pointing at. Attributed orders on Midwest Hardlines campaigns fell 45% with impressions flat and abandons up 1.5× — the spend is landing, the guests are arriving, and the product is not there. Four Midwest stores in the bottom-quartile remodel cohort hit stockout on the top-three Toys SKUs by 11am: around 70 stockout events on the day against a daily baseline near zero. The demand did not evaporate — marketplace Toys orders in the same region ran 85% above baseline as guests bought the same thing from third-party sellers.
At a glance
What moved
The four signals, on one morning
Impressions on plan is what rules out the media. Orders down with abandons up is the guest reaching the page and finding nothing. Marketplace up 85% in the same region is where they went instead — three datasets, one morning, no join written by hand.
The cause against the effect · stockout events and paid orders, day by day
The cohort runs elevated across the Black Friday window and spikes on the day itself. The paid line only breaks on the 28th — which is the day the top-three SKUs actually ran out, not the day the campaign changed.
Why · the stockout decomposed, one value per dimension
Contribution analysis of the stockout events. Markdown dollars on the same cohort ran at 0.40× baseline — there was nothing left to discount, which is the corroborating signal that this is an inventory event and not a pricing one.
Why the daily read is the whole value here
A weekly roll-up shows a good Black Friday. Read on the day, the three signals arrive together inside a window where the stores can still reallocate and the bids can still be paused — which is the only window in which any of this is actionable.
So what
Two actions, both inside the same day. Pause the Hardlines and Toys conversion bids in the Midwest until the shelf is back, because every impression until then is buying an abandon. And reallocate from the distribution centre to the four stores now — the demand is provably still there, it is simply being served by third-party sellers instead.
The points that matter
Flat impressions is what makes it diagnosable
Delivery held while orders fell 45%. That single pairing rules out budget, pacing and creative and leaves what the ad is pointing at.
−45%on flat deliveryThe demand is still there
Marketplace Toys orders in the same region ran 85% above baseline on the same day. The guest still bought — from a third-party seller, at a lower take rate.
+85%3P ordersIt is four stores, not a chain problem
74% of the stockout is Toys, 58% Midwest, and it sits in the bottom-quartile remodel cohort. The fix is an allocation call on a named handful of stores.
74%one categoryRecommendation
Pause conversion-objective bids on Hardlines and Toys in the Midwest until replenishment lands, issue a same-hour DC reallocation to the four stores, and check whether the bottom-quartile remodel cohort should have had the pre-Black-Friday allocation lift the top-quartile cohort received.
Questions it already answers
Is this a media problem?
No. Impressions were delivered on plan all morning; attributed orders fell 45% and abandons rose 1.5×. The campaigns worked — the shelf behind them did not.
How much of the lost demand did we actually lose?
Less than the paid numbers suggest. Marketplace Toys orders and GMV in the Midwest both ran 85% above baseline on the day, so a large part of it converted at third-party take rather than at owned-brand margin.
Why did markdown dollars fall on a day like this?
Because there was nothing to mark down — the cohort ran at 0.40× its normal markdown. That is the corroboration: the units are gone, not unsold.