Consumer Confidence Just Cratered. In July, Nobody Told Their Wallet.

Two numbers, nine days apart, telling what sounds like the same story.

August 14: headline retail sales fell 0.6% in July. A few days into August: the University of Michigan's consumer sentiment index posted one of its steepest single-month drops in years, missing every forecast on the board.

Falling spending, falling confidence — case closed, right? Open the actual Census tables and the picture splits in half. Some of what people buy from small operators went up in July. What collapsed was concentrated almost entirely somewhere else.

The headline number is doing a lot of hiding¶

Census reported it plainly: "retail and food services sales for July 2026... were $763.6 billion, down 0.6 percent... from the previous month." That's the number every wire service led with, and it's accurate.

It's also aggregated across categories that have almost nothing to do with each other, and this month they moved in opposite directions.

Two categories — cars and online retail — fell hard enough to drag the whole headline negative. Neither is where most small, local operators compete. Meanwhile food services rose half a percent, clothing rose nearly two, general merchandise and building materials both rose. Those are Main Street categories, and in July they were not where the trouble was.

Then sentiment fell off a cliff anyway¶

Here's where it stops being a tidy "actually things are fine" story.

The University of Michigan's Surveys of Consumers reported a preliminary August Index of Consumer Sentiment of 51.0, down from July's final reading of 55.2, against a consensus estimate of 54.5. The Index of Current Economic Conditions came in at 51.8; the Index of Consumer Expectations at 50.6. Both missed.

That last figure is the one worth sitting with. The survey's own director's statement puts it plainly: only 8% of consumers now expect their income growth to outpace inflation in the year ahead — down from 18% in December 2024. Year-ahead inflation expectations in the same release sit at 4.3%; long-run expectations at 3.3%.

People are not confused about their situation. They are telling a survey, in plain terms, that they expect to fall further behind — while, in the same month the sentiment data was collected, still going out for food, still buying clothes, still shopping general merchandise at rates roughly flat to slightly up.

Two different instruments, measuring two different things¶

This isn't a contradiction so much as a reminder that "the consumer" is measured by two genuinely different tools, and they answer different questions.

Retail sales measures what already happened. Actual dollars, actual transactions, recorded after the fact. Sentiment measures what people expect and how they feel about it, collected by asking them directly, before it necessarily shows up in a receipt.

Spending is often the last thing to move. A household that's newly anxious about the year ahead doesn't cancel dinner out the same week — it keeps behaving normally for a while, then starts trimming, usually starting with the things that are easiest to defer: a new car, a big-ticket online order. That is almost exactly the shape of July's data: the deferrable, big-ticket, non-local categories fell; the recurring, local, harder-to-defer categories held.

What to actually do with a split signal¶

1. 1

   Check which side of the split your own category sits on

   Motor vehicles and online-heavy nonstore retail did the damage; food service, apparel, general merchandise, and building materials did not. If you're in the categories that held, don't read the headline as your headline — read the breakdown as yours instead. If you're adjacent to autos or compete directly with large online retailers, the headline number is closer to your actual exposure.

2. 2

   Watch your own transaction count against your own ticket size

   This desk flagged the same divergence in the national data back in early August: revenue holding while visit counts fall is a different, more fragile situation than revenue holding because more people are coming in. Run the same split on your own numbers before assuming July's category data means you're fine.

3. 3

   Treat the sentiment drop as an early warning, not background noise

   An 8%-expect-a-real-raise reading is a specific, quotable, low number, not vague pessimism. If your customers are the ones telling that survey they expect to fall behind, plan for spending to follow sentiment down with a lag — price and staff for that now rather than reacting once it shows up in your own receipts.

4. 4

   Don't extrapolate one month of category data into a trend

   One month of retail sales by category is a data point, not a pattern — and this desk has already covered a release this month that was revised significantly after its first print. Note the split, watch whether it holds in August's data, and avoid treating July as destiny either direction.

Frequently asked questions¶

The lag is the whole story¶

Here's what I keep returning to: nothing in this piece is actually a contradiction. A household can genuinely believe things are about to get harder and still, this month, keep its normal routine — the coffee, the haircut, the new shirt. That's not denial. That's just how spending habits work; they're sticky, and they break slowly, usually starting with whatever's easiest to postpone.

Which means July's data isn't the reassurance it might look like at a glance, and it isn't the disaster the headline implied either. It's a snapshot of a lag — the gap between a household already telling a researcher it's worried, and that same household not yet changing what it does on a Tuesday.

The categories that held up this time are the ones worth watching hardest next.

It's just business — and confidence moves first.