The 2026 Holiday Forecasts, With the Population Attached to Every Number

In September two forecasters put a number on American holiday e-commerce for 2026. Adobe said $275.1 billion. Deloitte said at least $316.1 billion. Both figures are being quoted this month as the holiday e-commerce forecast, and they are $41 billion apart.

Neither is wrong. They are answers to different questions, and the questions are printed in the releases for anyone who reads past the headline. This page sets out what each forecast covers. The number that matters most to a loyalty program is in none of the headlines, and it comes last.

Published 4 November 2026; written 13 October 2026. Every figure was read on the publisher's own page and is linked. These are forecasts and predictions made before the season. Mastercard and Bain have also published 2026 outlooks; neither primary page would load while this was written, so neither is quoted.

The forecasts, side by side

Publisher	Published	What it covers	Period	Headline
Adobe	28 September 2026	US online sales, from Adobe Analytics transaction data	1 November to 31 December 2026	$275.1 billion, up 6.7%
Deloitte	10 September 2026	US retail sales, all channels, excluding automotive and gasoline	November 2026 to January 2027	$1.70 to $1.71 trillion, up 4.0% to 4.8%
Deloitte	10 September 2026	US e-commerce sales	November 2026 to January 2027	$316.1 to $318.9 billion, up 7.5% to 8.4%
Salesforce	20 July 2026	Global. Predictions about traffic and cost, with no sales total	"This holiday season"	20% of holiday ecommerce traffic from AI chat agents; $3 billion more spent subsidizing free shipping
NRF	Not published by 13 October 2026	US retail sales	November and December	None yet. Its 2025 forecast was 3.7% to 4.2%

Why $275.1 billion and $316.1 billion are both right

Adobe's figure is for "the period from Nov. 1 to Dec. 31, 2026" and is built from what Adobe Analytics observes: the press release announcing that figure cites more than one trillion visits to US retail sites and 100 million SKUs. It is online only and US only.

Deloitte's season, in its press release forecasting holiday retail sales, is "the November 2026 through January 2027 period". Its e-commerce range of $316.1 billion to $318.9 billion therefore includes a month Adobe leaves out, and its comparison base is the $294 billion it reports for last season from US Census Bureau data.

Two definitions of a season and two sources of data produce two totals. A third forecaster with a third definition will produce a third, and none of that is in a merchant's hands. What a merchant controls is whether the label travels with the number when it is pasted into a planning deck.

What Salesforce published is something else

Salesforce's 2026 holiday piece is titled as predictions and dated 20 July. Its lines are of this kind: "20% of all 2026 holiday ecommerce traffic will originate from AI chat agents", and "Retailers will spend an additional $3 billion globally to subsidize free shipping this holiday season".

Those are statements about traffic and cost, at global scale, made four months before Black Friday. They are worth reading. They are not interchangeable with a sales forecast, and the 20% in particular counts traffic that Salesforce itself describes as including "competitor scrapers" alongside shoppers.

NRF, as of 13 October

The National Retail Federation's holiday forecast is the one this table is missing. When this page was written NRF's own forecasts page carried two items: the 2026 annual outlook, in which "retail sales in 2026 will grow by 4.4% over 2025 to $5.6 trillion", and last year's holiday forecast, which predicted that "retail sales in November and December will grow between 3.7% to 4.2% over 2024".

That gives the shape to expect. NRF defines the season as November and December, the same two months as Adobe, across retail and not only online. If its 2026 figure has appeared by the time you read this, that is where it belongs in the table above.

The lines a loyalty program should read

Below the headline, Adobe's release has four details that bear on how a retention program should spend the season.

Cyber Week is 17.3% of the season

Adobe's holiday forecast expects Cyber Week, the five days from Thanksgiving to Cyber Monday, to bring in $47.5 billion, "representing 17.3% of the full holiday season spend". The other $227.6 billion is spent on the 56 days around it.

A loyalty calendar that runs a multiplier from Thursday to Monday and then goes quiet is built for a sixth of the money. The same release puts October at $95.8 billion, up 8%, before the season it defines has started.

Discounts peak lower than the headline suggests

The "up to 30%" in the summary of Adobe's forecast release is the peak for one category. By category, it expects discounts to peak at 30% off listed price in electronics, 29% in toys, 23% in TVs, computers and apparel, 19% in sporting goods, and 18% in appliances and furniture.

Those are peaks across large US retailers, and a smaller brand may have reasons to go deeper. But a store planning 40% off is 17 points deeper than the apparel peak, and every loyalty cost that is calculated on a discounted order grows with the depth of the discount.

Rewards sites are an acquisition channel for somebody else

Adobe reports that within affiliates and partners, "cashback and rewards partners leads in affiliate traffic with a 36% share". A shopper who arrives through one of those is collecting a reward from the partner, on top of anything your own program offers her.

AI traffic is growing from a small base

Adobe "expects AI traffic to U.S. retail sites this holiday season (measured by shoppers clicking on a link) to rise 130% YoY". That is a growth rate for visits. The release gives no share of sales for it.

What none of this tells you

Not one of these numbers forecasts your store. A US total says nothing about a brand doing two million dollars in skincare, and a forecast that turns out right for the country can be wrong for every individual merchant in it. Treat the totals as weather.

How to carry a forecast

• Publisher and date of publication.
• Geography: US, or global.
• Channel: online only, or all retail.
• Period: two months, three months, or five days.
• Kind: a forecast of sales, a prediction about something else, or a survey of intentions.

If any of the five is missing, the number is not ready to be repeated. Two copying errors were already in circulation when this page was written, one giving Cyber Monday as $15.2 billion and one giving Deloitte's e-commerce floor as $361.1 billion. The releases say $15.1 billion and $316.1 billion.

How the two big days differ in practice is covered in Cyber Monday vs Black Friday; the loyalty mechanics for the weekend are in our BFCM loyalty strategies guide, and the wider plan in the Shopify BFCM guide.

The number that comes last

It is 23%, or whichever of Adobe's category peaks is closest to what you sell. Write your planned BFCM discount next to it. If yours is the deeper of the two, find out what it does to the cost of every point you issue that weekend before the sale goes live.

FAQs