Every August the same headline runs: back-to-school spending is up. One number, one season, one shopper. It's clean, easy to present, and describes almost nothing.

There is no single back-to-school shopper. There are at least two — buying different things, at different times, for different lives. Average them together and you get a figure that's technically accurate and practically useless: a portrait of a shopper who doesn't exist.

It hides the basket

For 2026, the National Retail Federation and Prosper Insights & Analytics put K-12 spending at a record $43.3 billion and college at a record $103.5 billion — college's first time over $100 billion. Same "season," different baskets: K-12 runs on electronics, clothing, shoes, and supplies; college on electronics, dorm furnishings, and — tellingly — food and personal care. A blended shopper buys both a lunchbox and a mini fridge. No real household does.

Watch the scope, too. Those totals are built up from planned per-household spend. eMarketer's widely cited $85.42 billion measures something else — projected back-to-school retail sales. Both are real; they count different things. Stack them in one sentence, as roundups do, and you've built an average from two rulers.

It hides seasonality

The two baskets don't even fill at the same time. MNTN's 2026 research found college shoppers skew late — 42% start three to four weeks before classes, 24% wait until the final one to two weeks — while K-12 families increasingly buy early, a quarter of them as far out as June. eMarketer's guidance was blunt: treat college as a separate campaign, because the season now stretches through September.

So the two clocks run weeks apart. K-12 front-loads into summer; college backs up against move-in. Blend them and the average peak lands between the two, in a dead zone. Nobody is shopping at it.

It hides the audience

Here's the part no survey will ever hand you. A college freshman at move-in isn't buying school supplies — they're stocking a life they're running for the first time. And it shows up in what lands in the cart.

Insight Cloud reads exactly that, through GoPuff's on-demand ordering. In orders during 2025 move-in weeks, college-town baskets over-indexed against family markets on the categories you'd expect from someone figuring out adulthood in real time: party supplies (about 2.1× the family rate), family planning and personal care, frozen comfort food led by ice cream, ready-to-eat deli, vaping, and — a week or two later — cold and flu. Not a backpack in sight.

Basket size tells the same story. College ran $31 against family's $49 — the signature of a run-now purchase. And it didn't stop at move-in. The ordering stayed elevated through September, long after the planned-shopping surveys had called the season over.

That's the shopper the average can't see, and the opening most brands miss. Back-to-school lifts far more than notebooks and backpacks. Without a view into the impulse shopper, no one can stock or sell against those categories.

Disaggregation is the job

The blended back-to-school number isn't wrong. It's just not built to answer the question you actually have. It hides the basket, seasonality and audience all at once.

Some of that a careful survey can recover — the separate baskets, the offset timing. But the person, the freshman buying party cups and cold medicine in small run-now baskets, only shows up when you read behavior as it happened. That work used to take weeks of stitching. Reading behavioral signals directly, it's a question you answer the same afternoon.

A survey that asks what you'll buy for school never comes back with party cups and cold medicine. Behavior does.

Want to pull your category's back-to-school number apart? Request a demo of Insight Cloud.

A note on the numbers: NRF's totals are built from planned per-household spend; eMarketer's figure is a retail-sales forecast; the GoPuff figures describe observed ordering behavior. Three distinct lenses — we've kept them separate rather than blending them (which is, after all, the point).

Sources: National Retail Federation and Prosper Insights & Analytics, 2026 Back-to-Class Survey (July 14, 2026); MNTN Research, "Back-to-School Isn't What It Used to Be: 3 Trends for 2026," May 2026; eMarketer, "FAQ on back-to-school marketing," June 26, 2026. Behavioral ordering patterns from GoPuff on-demand data via Insight Cloud by Seek, 2025 back-to-school season.