What consumer sentiment actually measures — and what it quietly skips
Consumer sentiment is one of the most quoted numbers in markets and one of the least inspected. A print jumps two points, a headline says households turned gloomy, and index futures twitch. The object behind that twitch is narrower than the language around it.
A sentiment survey asks people to rate their own finances, the business climate they perceive, and what they expect over a defined horizon. It does not observe receipts. It does not audit bank accounts. It does not know whether the respondent delayed a car purchase, switched to private brands, or simply answered while a utility bill sat on the table. The series is a structured opinion, repeated so that changes in the opinion can be compared over time.
Three questions hiding inside one index
Most headline confidence composites mix current conditions with expectations. That blend is useful as a dashboard and dangerous as a causal story. Current-condition questions move with employment, wages, and prices the household has already paid. Expectation questions move with news, election cycles, equity prices, and whatever the respondent believes about “the economy” as a character in the news.
When both legs rise, the composite is easy to narrate. When they split — conditions firm, expectations sour — the headline can hide the split. CrowdMood publishes the legs separately for that reason. A retailer staffing for the next quarter cares more about purchase intent than about a five-year business-outlook question that tracks presidential approval.
The sample is a product, not a mirror
Who is asked, how they are reached, and how non-response is weighted decides the print as much as mood does. Landline-era frames under-counted mobile-only adults. Online panels over-represent people who fill out surveys. Seasonal weighting can smooth a genuine shock. None of this makes surveys worthless. It makes them instruments. Instruments have drift.
Compare two famous U.S. series. The University of Michigan Surveys of Consumers grew from a face-to-face academic project into a widely watched monthly. The Conference Board’s Consumer Confidence Index uses a different question set and a different sample. They often rhyme and sometimes diverge for months. Divergence is information: it usually means the questions are loading on different household stresses, not that one series is “wrong.”
What surveys skip on purpose
Surveys are weak at high-frequency product substitution. They will tell you households feel squeezed. They will not tell you that olive oil was swapped for a cheaper fat, or that a streaming bundle was cancelled the week after a rent reset. They are also weak at the top of the income distribution, where a small number of households drive a large share of discretionary spend. A confidence index can look stable while luxury demand cracks, or look weak while warehouse clubs print record traffic.
They also skip credit conditions except as the respondent perceives them. A household with a thick FICO file and a household at a 24% APR card can give the same “good” rating to current finances if the first is complacent and the second has not yet missed a payment. Sentiment is not a credit model.
How to read a print without fooling yourself
Start with the question, not the nickname. “Confidence” and “sentiment” are marketing words for different questionnaires. Then look at the one-month change against the three-month average, not against a trader’s muscle memory of last year’s level. Then check whether the move is in current conditions or expectations. Then ask which sector should theoretically care: housing questions for lenders, vehicle questions for OEMs, vacation questions for airlines.
Finally, pair the survey with a hard series that cannot lie about volumes — card spend, retail sales, jobless claims, or our own sector tape. When mood and tickets travel together, you have a cycle. When they diverge, you have a research problem, which is more valuable than a tidy story.
CrowdMood’s public dashboard is an educational composite with a published recipe. It is not a substitute for official statistical agencies. It exists so that a global reader can see household mood as a dataset: dated, comparable, and explicit about uncertainty. If you license a signal from us, you are buying that discipline, not a promise that gloom predicts the next equity drawdown.