Margin & Markdown Sentinel · Wisdom notebook
Ask the follow-up.
Agentic conversational BI · one thread roaming the markdown book, cohort by cohort against each cohort's own baseline.
Cottagecore is already routed. What else in the book is discounting above its own seasonal baseline right now?
RKScored every department × banner × trend cohort's markdown dollars against its own seasonal baseline overnight, set aside the cohort already routed, and ranked what remains — in both directions, because a cohort discounting far below its baseline is a finding too.
One other cohort is genuinely hot, and it is not a trend cohort at all — owned-brand Beauty is running 2.4× its own markdown baseline through Q3. Everything else is at or under baseline, and two cohorts are notably under it:
Markdown dollars against each cohort's own seasonal baseline
Read as multiples of each cohort's own baseline, so a big department and a small one are comparable. National-brand Beauty sits at 0.30× over the same quarter — which is the other half of the owned-brand story, not a separate one.
Quiet Luxury at 0.39× — is that a cohort holding price, or just a quarter where nothing needed clearing?
RKPulled the Quiet_Luxury cohort's Q4 read against the rest of the apparel book on the four measures that separate a disciplined cohort from a quiet one: markdown share, return rate, forecast variance and the efficiency of the media pointed at it.
Holding price — and on every measure at once, which is what makes it a signature rather than a quiet quarter. Guests are buying it intentionally and keeping it:
Quiet_Luxury, Q4 FY25 · against the rest of the apparel book
A quiet quarter moves one of these. All four moving together — full price held, fewer returns, sales ahead of plan, and the ads against it running at nearly twice baseline efficiency — is the cohort, not the calendar. It is also the exact inverse of the Cottagecore launch, on the same forecast engine.
Beauty is discounting less in November while owned-brand Beauty discounted more in Q3. What changed?
RKRead November's Beauty department across all three datasets — sell-through against forecast, markdown dollars against baseline, retail-media yield by advertiser tier, and the marketplace take rate on the same category.
The media, not the price. Beauty sold 18% ahead of its seasonal forecast in November on 0.85× its normal markdown — the volume came from the campaigns, not from discounting. The interesting part is which advertisers delivered it:
Beauty · November 2025
The mid-spend beauty advertisers return 6.2× against the mass tier's 3.9× and are served 23 points less of the inventory they could have. Marketplace beauty take rate is up 20% in the same month, so the lift is incremental rather than shifted — which makes the impression cap, not the markdown budget, the thing worth moving.