U.S. shoppers will spend a record $275.1 billion online this holiday, Adobe Analytics forecasts — 6.7% growth on last year, and $95.8 billion of it in October alone [1]. A consumer study of more than 15,000 shoppers says those same people are buying fewer units and moving to cheaper stores [2]. My read: each forecast misses what the other counts, and the series a plan has to ship is the one falling.

The big picture:

The gap is about what each one counts. Adobe tracks the dollars leaving online checkouts [1]. The consultancy AlixPartners tracks behaviour: basket size, the discount a shopper waits for, and which stores get the trip [2].

Retail plans run two books. The top is money — open-to-buy, the budget a buyer may commit, and the financial plan behind it [3]. The bottom is physical, where restocking and allocation place units in stores. When dollars rise while items fall, one merged plan cannot serve both.

The calendar shifted too. October now carries $95.8 billion of online spend, up 8% on last year, and the five days from Thanksgiving to Cyber Monday hold 17.3% of the whole season [1]. Adobe’s own note says that early rush “has changed the shape of the holiday season” [1].

Shoppers are funding the rush with deals and credit. AlixPartners found 56% act only on discounts of 30% or more, and two in three expect at least a third of what they buy to be on deal [3]. Card balances hit $1.26 trillion in the second quarter of 2026, up 5.9% and rising faster than sales, while payments more than 90 days late reached 12.8%, the highest since 2011 [3].

What strikes me is the split between mood and behaviour inside the same research. Fifty-seven percent say the economy is worse than a year ago and 37% plan to spend less, yet spending has beaten last year by about 5% in every month of 2026 [2].

By the numbers

  • 6.7% — Online holiday dollars: Adobe expects a record $275.1 billion of U.S. online sales from Nov. 1 to Dec. 31, with early deals carrying part of the gain [1].
  • $95.8 billion — October pull-forward: Online spend in October alone, up 8% on last year, including $9.9 billion over Prime Day on Oct. 6 and 7 [1].
  • 30% — The discount bar: 56% of shoppers act only on deals of 30% or more; Adobe puts the deepest Cyber Week cuts at up to 30% off [1][3].
  • $1.26 trillion — Credit balances: Card balances in the second quarter of 2026, up 5.9% on a year earlier, with late payments past 90 days at 12.8% [3].

What I’d watch:

Retailers are answering by merging the two books. On Oct. 8 the software firm Relex Solutions said Neighborhood Intelligence — the owner of Bed Bath & Beyond, The Container Store and other home banners — will run forecasting, restocking and allocation beside the money side of the plan on one data layer across 335 stores [4]. The vendor says the first tools went live “in a matter of weeks” [4].

Gartner’s Sept. 16 research on warehouse Artificial Intelligence (AI) leads its list with demand forecasting, praising the “transparency and repeatability” older methods already had [6]. The consultancy’s own advice points the same way, urging retailers to put AI into demand forecasting rather than discounting [3].

Capacity still moves on a slower clock. Tractor Supply opened a $200 million, 865,000-square-foot depot in Nampa, Idaho, built from a design drawn up in December 2024 and shipping to 123 stores across nine states from October [5].

The part I keep circling is the measuring. In a May survey of 661 retail and consumer-goods stock professionals, 51% said their teams at times go ahead with demand forecasts they know are shaky because no better option exists, and 16% said they do not track forecast accuracy at all [7]. For that sixth of the market, a better model is not the bottleneck.

  • The open-to-buy release: the mid-season point where a money budget turns into unit orders, and the two series have to agree [3].
  • The discount assumption: if only deals of 30% or more move goods, the discount line has become the demand plan rather than a lever on margin [3].
  • The markdown trigger: the date a plan decides unsold stock is worth less than its shelf space, which sets whether the miss lands this quarter or the next [7].

The catch

The clearest limit is what the unit data covers. The fall in units comes from receipts between January and July, not from the holiday quarter, and Adobe’s number is a guess built from past transactions rather than a result [1][3]. Two different groups are being compared, and neither is a scorecard for December.

I could be wrong to read this as a unit problem when it may still be a price one. If rising prices and credit keep the dollar figure climbing, a plan can look healthy through a shrinking basket [3]. That ends when the markdowns land, because a 30% discount bar then stops being a story about sales and becomes one about margin.

None of this argues that forecasting tools are wasted. It argues that the basis of the plan — the unit and the week — is the part that moved, and no model can square two series a plan treats as one.

At a glance

  • The Big Shift: Adobe forecasts a record $275.1 billion of U.S. online holiday sales, up 6.7%, with $95.8 billion in October alone [1]. Research on the same season finds shoppers buying fewer units and waiting for discounts of 30% or more [2].
  • Why It Matters: Retail plans hold dollars at the top, in open-to-buy and the financial plan, and units at the bottom, in restocking and allocation [3]. When dollars rise while shipped items fall, one merged plan hides the gap that reaches the shelf.
  • What I’d Watch: Whether the mid-season plan gets re-cut in units instead of dollars.
  • The open-to-buy release: the point where a money budget becomes unit orders, and the two series have to agree.
  • The discount bar: the depth of deal that actually moves goods, now an input to demand rather than a lever on margin.
  • The markdown trigger: when unsold stock is written down, which decides whether the miss lands in this quarter or the next.
  • The Catch: The unit decline is measured on receipts through July, and Adobe’s number is a forecast rather than a result [1][3]. A dollar plan can look healthy while the basket shrinks, up to the moment the markdowns land.

Related reading

Sources

[1] Adobe — “U.S. Holiday Shopping Season to Hit Record $275.1 Billion Online, Rising 6.7% YoY” (September 28, 2026) — https://news.adobe.com/news/2026/09/adobe-us-holiday-shopping-season-to-hit-record [2] AlixPartners — 2026 U.S. Retail Holiday Outlook press release (September 16, 2026) — https://www.alixpartners.com/newsroom/press-release-us-holiday-sales-expected-to-grow-4-7-this-season-even-as-consumer-confidence-declines [3] AlixPartners — 2026 U.S. Retail Holiday Outlook Special Report (September 16, 2026) — https://www.alixpartners.com/media/14sbg14k/2026-us-retail-holiday-outlook-special-report__vf.pdf [4] Relex Solutions — “Neighborhood Intelligence to Streamline Supply Chain and Merchandise Planning with Relex” (October 8, 2026) — https://www.relexsolutions.com/news/neighborhood-intelligence-to-streamline-supply-chain-and-merchandise-planning-with-relex [5] Tractor Supply Company — “Tractor Supply Celebrates Grand Opening of 11th Distribution Center in Nampa, Idaho” (September 23, 2026) — https://corporate.tractorsupply.com/newsroom/news-releases/news-releases-details/2026/Tractor-Supply-Celebrates-Grand-Opening-of-11th-Distribution-Center-in-Nampa-Idaho/default.aspx [6] Gartner — “Top Four AI Trends Transforming Warehousing for Supply Chain Leaders” (September 16, 2026) — https://www.gartner.com/en/newsroom/press-releases/2026-09-16-gartner-identifies-the-top-4-ai-trends-transforming-warehousing-for-supply-chain-leaders [7] Doss — “Shelf Life Stress: What’s Keeping Inventory Teams Up at Night in 2026?” (published May 18, 2026) — https://www.doss.com/research/top-inventory-stressors-affecting-teams