The back-to-school numbers for 2026 are the kind of numbers a merchant frames and hangs on the wall. K-12 spending reached a record, college crossed a threshold it had never crossed, and the average family opened its wallet a little wider than it did a year ago. If you run a retail P&L, that reads as validation: the consumer showed up, the assortment worked, the season closed strong. I want to make the case that the record is real and the interpretation is dangerous, and that the gap between those two things is exactly where retailers will lose margin in the fourth quarter of 2026.
A record built on hedging, not on conviction
Start with the topline, because it is genuinely a record. The National Retail Federation and Prosper Insights & Analytics put K-12 back-to-school spending at $43.3 billion for 2026, up from $39.4 billion in 2025 and above the previous high of $41.5 billion set in 2023. College spending hit a record $103.5 billion, going past $100 billion for the first time and jumping from $88.8 billion the year before. The average K-12 family planned to spend $863.86, a step up from $858.07 in 2025. Those are the figures that make it into the earnings script.
Now look at how the record was assembled, because the composition tells a different story than the total. By early July 2026, 62% of shoppers had already started, which the survey itself frames as a decline from 67% at the same point in 2025 and a rise from 55% in 2024. More than half of shoppers, 54%, told NRF they bought school-related items during the June sales events: Prime Day, Walmart Deals, and Target Circle Deal Days. Among shoppers who had not yet finished at least half their list, 46% said they were deliberately waiting for the best deals, and 23% said they were spreading their budgets out over time. Roughly half, 47%, planned to buy only the essentials for the first day and replenish later. And 78% of consumers said they anticipated higher prices on back-to-school items this year, with Deloitte separately forecasting the season to come in flat and 57% of consumers expecting the economy to worsen, the highest reading since 2020.
Put those two pictures side by side. A record total. A consumer who moved early to catch promotions, is holding back the rest of the list until prices break, is buying only what is strictly needed, and expects to pay more. That is the behavior of a household hedging against price uncertainty, and it happened to hedge in a year when enough families hedged at once to set a record.. The dollars are real. The conviction behind them is thin.
Why the signal reads as loyalty when it is really a hedge
Here is the trap, and it is a structural one. A retail forecasting system is built to reward two things: revenue and recency. It sees a category clear through faster than last year and it reads that as strong sell-through. It sees a record total and it reads that as demand. What it does not natively see is the difference between a basket that exists because a family finally decided to buy, and a basket that exists because a family that was going to buy anyway was pulled forward by a 25-cent loss leader or a rack of items priced below $25. Both baskets look identical in the data. Both close revenue. Only one of them tells you anything about underlying demand.
This is why the discounting escalated the way it did. Walmart marked select items to 25 cents and reset the prices on its fourteen most popular school supplies to their lowest levels since 2019. Kohl’s promoted thousands of products under $25. Target debuted its offers in June and refreshed more than half of that inventory. Each of those moves was rational in isolation, and together they describe a market where every major player decided the safest way to defend share was to bid the deal up earlier and deeper than the player next door. When 46% of the remaining shoppers are openly waiting for a better price, the promotion is a toll you pay to collect demand that has learned to wait.. The retailer is effectively paying to be told what it already suspected, which is that it cannot read its own customer without a discount attached.
That is the distrust I want to name. When you cannot separate the baskets you earned from the baskets you pulled, you stop trusting the signal, and once you stop trusting the signal you default to the one thing you know moves volume: price. The record spending figure then becomes cover. It lets the organization tell itself the promotions worked, because look, we set a record, when the more honest reading is that the promotions were the price of admission to a flat-to-modest year that a lot of households paid for in advance.
The evidence that this is timing and price, not renewed appetite
You do not have to take the behavioral read on faith, because the category detail confirms it. Electronics remained the largest K-12 line at $14.7 billion, and yet the average planned spend per student on electronics slipped to $293.11 from $295.81 the year before. That is the tell. The category is enormous, it is growing in aggregate, and the individual family is spending marginally less on it than last year. Growth in the total is coming from more households participating and from mix, not from any single family deciding the moment is right to trade up. Shoes rose about 2.9% per family to $174.01 and school supplies rose about 1.9% to $146.45, increases that track closely with the price pressure 78% of shoppers said they expected. When per-family spend rises roughly in line with expected price increases, you are watching families run to stand still. The record is the sum of a lot of families each covering a slightly more expensive list, front-loaded into promotional windows, with the discretionary tail deferred.
Channel behavior points the same direction. Online fell to 50% of K-12 shoppers from 55% a year earlier, and among college shoppers it dropped to 41%, the lowest since 2016. In a year defined by deal-hunting, shoppers moved back toward department stores at 47%, discount stores at 44%, and clothing stores at 39% for K-12. That is a consumer physically working the promotions across formats, comparing shelf prices, and buying where the markdown is. It is deliberate, price-led shopping. It is a signal that they want your lowest price and will drive to three formats to find it..
So the season that looks like appetite is, on inspection, three things stacked on top of each other: earlier timing, driven by June promotions; per-family spend rising with prices; and a firm cap on discretionary purchases, with 47% buying essentials only and the rest waiting. None of those three is loyalty. All three are consistent with a household managing a budget it does not trust the economy to protect.
What the misread costs when it flows into Q4
Now the expensive part. The back-to-school read becomes an input to holiday planning, and this is where booking the record as good news turns into real dollars of shrink and markdown..
If your forecast treats the fast August sell-through as a demand signal rather than a timing signal, it will lift the buy on exactly the categories that cleared, and it will lift them into a fourth quarter when the same households are still hedging. You will over-order the SKUs that only moved because they were the loss leaders, and you will under-appreciate that the discretionary tail those families deferred in August is the same tail they will defer again in November if prices do not break. The record becomes a multiplier applied to the wrong base.
The mechanism compounds through your own promotional calendar. Having taught the customer that the real price arrives with the deal, you now inherit a Q4 consumer who has been trained across an entire back-to-school season to wait. The 46% who held out for a better price in July carry it into holiday, which means your promotions in November are competing against a customer who has structural patience you built.. If you forecast holiday off an unadjusted back-to-school record, you will over-buy, then you will have to promote deeper than planned to clear it, and the margin you already conceded to pull the August basket forward you will concede again to move the December inventory. That is paying the toll twice on the same demand.
And the inventory concentrates in the categories where the promotional intensity was highest and the per-unit economics were thinnest, which are precisely the ones the flat-appetite signal was warning you about.. Electronics is the clearest case: the biggest category by dollars, growing on participation, shrinking slightly per family. Buy that on a growth read and you are long the most capital-intensive, most markdown-exposed category going into the exact quarter when the household told you, at 57%, that it expects things to get worse.
The operating model that separates pulled demand from real demand
The fix is a decomposition step inserted before any back-to-school number is allowed to inform a Q4 buy.. Three moves make it work.
First, decompose every category’s result into incremental baskets versus pulled-forward baskets before you book it. The instrumentation already exists in most enterprise retail stacks: promotion flags, timing of purchase relative to the June sales windows, and repeat-versus-new cohort tags. Run the sell-through against the promotional calendar and ask a single question of each unit of growth, which is whether it would have existed at the shelf price in the same week absent the deal. The units that would not have are timing, and they get discounted out of the demand base that feeds holiday. What remains is the number you can trust.
Second, treat the deferred tail as a live liability rather than upside. When 47% of families buy essentials only and 46% are waiting for price, you are carrying a known quantity of demand that is conditional on future markdowns. Model it as a conditional forecast, priced at the discount the customer is clearly holding out for, and plan the Q4 inventory and margin around that condition instead of assuming the tail converts at full price. This is the difference between planning for the customer you have and planning for the customer the record implies.
Third, put a distrust check on the promotion itself. Before extending a deal, measure whether it is generating incremental demand or simply harvesting demand that would have arrived anyway. If the June and July data show that the deal mostly moved the timing of a purchase without lifting the total, that is your signal that you are paying a toll, and the Q4 answer is a narrower, more targeted promotional plan rather than a broader one. The goal is to stop using price as a substitute for a demand signal you have decided you cannot read, and to rebuild a signal you can.
Retailers who run this decomposition will walk into holiday 2026 with a smaller, cleaner buy in the categories that only cleared on price, a funded plan for the discretionary tail the consumer deferred, and a promotional calendar sized to demand they actually earned. Retailers who book the $43.3 billion as validation will walk in long the wrong SKUs, promoting to clear them, and calling the second markdown a strong season again. The record was true. The question that decides the fourth quarter is whether you read it as a customer who came back, or a customer who was managing risk and happened to spend a record amount doing it. Plan for the second one, and the year still works.