The moment a news anchor says "grocery inflation has eased to around 2% year over year," someone standing at the checkout stares at their receipt and thinks: this is 2%? Neither is lying. Grocery CPI really is running near +2% versus a year ago, and the cart really does feel heavy. The catch sits elsewhere: the two numbers measure entirely different things. The statistic asks "how much higher than this time last year," while memory asks "how much more expensive than before the pandemic — than 2020." Same grocery aisle, yet prices jumped a cumulative +24.68% from 2020 to 2024. Different reference points, different conclusions.

Whose basket is CPI, anyway?

The Consumer Price Index (CPI) tracks the price of a fixed representative basket of goods. Hundreds of items — food, clothing, housing, transport, communications, education, healthcare — each carry a weight, and the weighted average produces a single number. Housing might be about 30%, food 15%, energy 8%.

Chew on two words here: 'weighted average' and 'average household.' Items that loom large in the average household's spending get high weights; rarely bought items get low ones. But how many people actually consume like the statistical average? Homeowners versus renters, car commuters versus transit riders, a family with school-age kids versus a single-person household — their baskets are completely different. The further your spending mix sits from the average, the further your perceived inflation drifts from official CPI. Simple, but nearly every gap starts right here.

Same 3% CPI, four households, four faces

One month's CPI reads 3%. The face that meets that 3% changes by household.

• Homeowner, mid-level spending — shielded from rent hikes, so it feels lower than CPI.
• Urban renter, car commuter — when rent and fuel rise together, "3%? My wallet is up 7%."
• Household with school-age kids — education and childcare spikes hit hard, while falling electronics prices barely register.
• Retired older household — heavy on healthcare and food; when those rise, they can feel more than double the CPI.

Why does the same index split into four faces? Because each wallet carries a different basket.

When the things you buy often go up, it hurts twice

The second reason lives in our heads. People don't remember every price with equal weight; prices we meet often imprint harder. Food, dining out, and transport are daily or weekly encounters, so even a small bump replays in your head as "up again." A price drop on furniture or appliances bought once a year? You barely notice it.

• Frequency effect — frequently bought items are over-represented in perception.
• Salience effect — visible, large hikes dominate memory.
• Loss asymmetry — we vividly recall milk rising 20 cents but forget seasonal produce falling.

Yet CPI weights items by average spending share, not by how often you buy them. Even if your daily instant-noodle price rises, a small spending share means a small index impact. The mental weight and the statistical weight diverge precisely here.

Break it down by category

The single headline CPI averages everything — rising items, falling items, slow movers. So a drop in one category quietly masks a surge in another. Even in the same period, the categories that drive perception and the ones that suppress the index move on separate tracks.

Housing is the clearest case. Buying a home is classified as an 'asset investment,' not consumption, so CPI captures shelter mainly through actual rents rather than home prices. When prices surge but rents move slowly, the index looks calm. Now look back at the table. Food rises just +2.2% yet feels large because you meet it daily, while auto insurance spikes +22.6% but barely sticks because you renew once a year. The size of the increase and how often you notice it pull in opposite directions.

You compare to '2020,' not to 'last year'

The decisive gap comes from the reference point. Statistics almost always report year-over-year. So once a price jumps and then holds, the rate naturally eases toward 0% the following year. The price didn't fall. It merely stopped climbing — and the statistic reads "stable."

Memory, however, grabs onto pre-pandemic 2020 prices as its anchor. By that yardstick, food sits up a cumulative +24.68% from 2020 to 2024. The rate may slow, but the price level stays parked far higher. The statistic watches 'speed' while people watch 'position,' so "they say it eased, but why is it this expensive?" keeps repeating.

Which index you read matters too. Measures that more broadly include living costs such as loan interest tend to print higher than core CPI. Even for the same prices, what goes in the basket changes the number.

✍️ Editor's note — Honestly, watching a "inflation has eased" headline and then getting hit at the register is practically a rite of passage — and the culprit is the reference point. Stats look at last year; our heads look at 2020. A cumulative +24.68% on food means it climbed nearly a quarter in four years, so quoting just the +2% year-over-year and calling it "stable" is bound to feel off. Both are true; they just measure different things. Instead of accusing the statistics of lying, it's easier to think, "ah, it's watching speed and I'm watching position."

Nominal vs real — the pair you must read together

To read perception accurately, pair it with your income. Your pay rose 3% but prices rose 5%? Your nominal wage went up while your real wage shrank — you can buy less at the store. That's how a "record-high salary" headline can mislead. The longer the comparison, the wider this gap. The average US household income in 1980 was about $22,000, which in today's purchasing power is roughly $78,000. Nominally that's more than triple; in real terms the change is far smaller.

Caveats and limits

The category figures above (food +2.2%, housing +4.5%, auto insurance +22.6%, cumulative +24.68%) are illustrative of a specific period and source, and vary by time, country, and agency. That doesn't make CPI a 'wrong number.' Tracking the macro trend and running monetary policy requires an average gauge, and statistics agencies survey tens of thousands of prices monthly to build it. At the same time, your hotter perception isn't an illusion — it's a rational response. Neither number is lying; they measure different things. This article is educational and not financial advice; every household's situation differs.

Intuition comes from repetition

Want to see exactly where perception and the official index split? The fastest path is to record the prices of items you buy often and then test your guesses. Use PriceGuess's grocery quiz to compare the same product's perceived price across countries and dates, and Time Machine to trace how a fixed sum changed to today. To see inflation inside a wider price sense, continue to crypto price guessing and feel, by hand, how a single headline number unfolds into real everyday prices.