A discount ceiling is not a discount target. Adobe's holiday forecast puts $275.1 billion of US online spending between November 1 and December 31, up 6.7 percent, and attaches seasonal ceilings of 14, 21 and 30 percent off listed price; as of November 10, 2026, many teams were reading those ceilings as targets rather than as boundsNoibu. The total is the headline; the distribution inside it is what changes staffing, pricing and page design decisions, because a season that grows 6.7 percent overall can still shrink for any single category, and a promotion priced against the ceiling instead of the reference price tends to fund itself out of margin.
Where the money is actually landing
Cyber Week carries $47.5 billion, up 7.4 percent, with Cyber Monday at $15.1 billion and Black Friday at $12.9 billion17TRACK. Five days absorbing roughly one sixth of the season is not a marketing fact, it is an operations fact: the same warehouse, the same support desk and the same payment stack have to perform at peak for a week and then return to normal.
The single-day growth rates differ in a way that matters. Black Friday is forecast to grow 9.2 percent while Cyber Monday grows more slowly, which suggests discounting is shifting back toward the store-and-app combination rather than the pure desktop workday purchase.
Mobile share is a majority problem
Mobile at 57.4 percent of the season, up from 56.4 percent a year earlier, means the majority of revenue passes through a phone screenNoibu. A friction that barely registers on desktop — a wallet button that fails to load, a keyboard covering a field — becomes a majority problem because it sits on the path most shoppers take.
The comparison worth running is mobile conversion against desktop conversion on the same templates. A wide gap is not a design opinion, it is a list of pages to inspect, starting with the ones that carry the most traffic during the five peak days.
The market size is not in dispute, the split is
Forecast totals rarely move much once a season starts, but the split between categories does. Consumer demand shifting toward replacements and consumables rather than big-ticket gifts changes the assortment a retailer should feature, and it changes which product pages deserve the traffic the forecast implies.
Discount depth has a ceiling, and it moves
| Period | Forecast peak discount | What it applies to |
|---|---|---|
| Early November | Up to 14 percent | Seasonal ceiling off listed price |
| Just before Thanksgiving | Up to 21 percent | Seasonal ceiling off listed price |
| Cyber Week | Up to 30 percent | Seasonal ceiling off listed price |
These are ceilings, not averages, and they do not apply to every category or merchant17TRACK. The practical use is as a sanity check: a promotion that claims to beat the seasonal ceiling without explaining the reference price is a promotion that will be compared against a price the shopper cannot verify.
AI referrals and the traffic nobody can see yet
Adobe expects traffic from AI tools to US retail sites to rise 130 percent this season, while total automated traffic already makes up the majority of global web trafficDevPro Journal. Two things follow. First, product pages have to be readable by both shoppers and agents. Second, bot filtering has to distinguish a genuine assistant from card testing and inventory hoarding.
Most teams cannot yet separate AI-referred sessions from the rest. A reasonable starting point is a simple segment: sessions with a referral from a known assistant domain, measured on conversion and return rate rather than on traffic volume. Volume alone will flatter the channel.
What to fix in the next two weeks
- Test checkout on real phonesEvery payment method, including buy now pay later, which Adobe expects to drive $21.3 billion this season. Test on the devices your traffic actually uses.
- Freeze releases before the peakA code freeze from the week before Thanksgiving through Cyber Monday keeps risky changes off the busiest days. Keep an emergency path for security patches.
- Separate estimates from guaranteesDelivery dates at checkout, on tracking pages and in support scripts have to say the same thing, including what happens when the date slips.
- Segment AI referralsTrack sessions from assistant domains separately. Judge them on conversion and returns, not on visit counts.
None of these requires a new platform. They require deciding, before the traffic arrives, which of the five days you are actually planning for and what a failure on each of them would cost. The forecast is useful only to the extent it changes what happens in the next fortnight.
FAQ
Is a 6.7 percent growth forecast good or bad?
It is close to inflation plus modest volume growth, which means the season rewards share gains rather than category tailwinds. Planning for growth without a share argument is planning for a flat result.
Why does the mobile share matter more than the total?
Because 57.4 percent means most revenue passes through the least forgiving interface. Small checkout frictions cost more there than anywhere else in the funnel.
Should AI referral traffic change the content plan?
It should change how product pages are written. Specifications, prices and offer conditions need to be explicit and consistent, because both shoppers and assistants read them as facts.
Are the discount ceilings a target?
No. They are seasonal upper bounds off listed price. Treating a ceiling as a target usually means funding a discount that the reference price cannot support.
