Last updated: 18 September 2026
Quick Answer: Calculate market size by multiplying a counted population by what each buyer spends per year. Build it twice, top-down from a published statistic such as the Bureau of Labor Statistics consumer expenditure tables, and bottom-up from your own penetration and price assumptions. A gap over about three times means a definition is wrong, not an estimate.
Market size is one multiplication repeated at three levels, and almost nobody gets it wrong by multiplying badly. The Bureau of Labor Statistics reports 135,760 thousand US consumer units in 2024 spending $978 a year on personal care products and services, which multiplies out to a $132.8 billion category from one public table.
The trouble starts next. A consumer unit is not a household, and that $978 line mixes shampoo with haircuts. Sizing errors are definitional far more often than arithmetic, which is why the same market gets built twice and published as a range.
Which Public Data Sets Give You Each Input?
Finding a market size without buying a report means assembling it from named data sets whose exclusions you can read. The table below is ordered alphabetically by source.
| Source | What it gives you | What it does not cover |
|---|---|---|
| Bureau of Economic Analysis, Personal Consumption Expenditures | Consumer spending by product category | Granular categories BEA itself flags as less reliable |
| Bureau of Labor Statistics, Consumer Expenditure Survey | Spend per consumer unit by category and income quintile | Households as such; income is under-reported |
| Census Bureau, County Business Patterns | Establishment counts by industry and geography | Businesses with no paid employees |
The Bureau of Economic Analysis says its detail tables rest more often on judgmental trends or weaker source data than its summary tables, and it publishes how far its first reading moves: from 1999 to 2022 its advance estimates of real consumer spending were revised by 1.08 percentage points on average, ignoring sign.
The Census row is wider than it reads: employer businesses were 21.6 percent of all US businesses in 2023, down from 24.6 percent in 2012, so County Business Patterns sees one business in five. It arrives late too: the Census Bureau published its 2023 Nonemployer Statistics on 15 May 2025, seventeen months after that year closed.
Two Worked Calculations of the Same Market
Take a premium refillable personal care product sold to US consumers at $12 a refill.
Top-Down, Source by Source
- Population. 135,760 thousand consumer units in 2024, from the Consumer Expenditure Survey. The Bureau of Labor Statistics defines a consumer unit by financial interdependence, not shared address, so multiplying it by a Census household figure mixes two denominators.
- Spend per unit. $978 a year on personal care products and services, same source, same year. A survey estimate, not a count: the Bureau of Labor Statistics puts the 2019 Consumer Expenditure Interview Survey response rate at 53.7 percent.
- Category total. 135.76 million multiplied by $978 gives $132.8 billion.
- Narrow to products. That line includes salon services and the split is not published, so this is an assumption, not a statistic. Assume products are 40 to 60 percent, giving $53 billion to $80 billion.
- Narrow to your segment. Premium refillable formats are a small slice. Assume 1 percent, and the addressable category lands at $530 million to $800 million.
Publish the range, never $665 million: the midpoint implies a precision steps four and five cannot support. Narrowing well is a segmentation problem before it is an arithmetic one.
Bottom-Up, Assumption by Assumption
Bottom-up starts from countable units, keeps every assumption in the open, and is the better method for a new product.
- Addressable units. The Consumer Expenditure Survey publishes by income quintile, so the top two quintiles are 40 percent of 135,760 thousand, or 54.3 million consumer units.
- Penetration. The share of those units that would buy a refillable format at a premium. Assume 15 percent, giving 8.15 million buyers. This is the number purchase intent research is good at estimating, and the one teams guess instead.
- Frequency and price. Six refills a year at $12 gives $72 per buyer per year, and what customers will actually pay rarely matches the price on the plan.
- Result. 8.15 million buyers multiplied by $72 gives roughly $587 million a year.
Now move the assumption doing the most work. Holding frequency and price, penetration alone swings the answer from $313 million at 8 percent to $977 million at 25 percent. That spread is why the output is a range, and a wider one than instinct suggests: Haran, Moore and Morewedge found stated 90 percent intervals held the right answer 74 percent of the time.
What Do You Do When the Two Methods Disagree?
Top-down gave $530 million to $800 million and bottom-up gave $587 million, so the two overlap. That is agreement: intervals from independent assumptions, not identical numbers. When they do not, the gap tells you which error you are hunting.
- Under about three times apart. An estimation error: an assumption is off, and the fix is to measure the one doing the most work.
- More than about three times apart. A definitional error: the two methods are sizing different markets. Check the units, then the category boundary, then the time period. A per-month price multiplied by an annual population is the commonest version.
- Bottom-up far smaller than top-down. Usually correct for a new product. The gap is expansion headroom, not an error to correct.
Who Is Missing From the Penetration Estimate?
The assumption you most need to measure is the one most likely to be measured on the wrong people. Penetration usually comes off an online panel, and Pew Research Center reported in 2025 that 16 percent of US adults are smartphone-only, running at 34 percent in households under $30,000 against 4 percent above $100,000. That suits a premium product. For a value-priced one penetration reads high, and the American Association for Public Opinion Research calls that coverage error rather than sampling error, which no sample size repairs.
The cheap fix weights the panel and widens the interval. The expensive one changes the mode so people a panel never sees can answer. Alchemic runs end-to-end consumer research at scale with WhatsApp-native interviews that need no link and no app, AI phone interviews, and recruitment as managed fieldwork or bring your own across 14 markets including the USA and the UK. Mode decides who ends up in the sample, and reaching respondents who are not on a browser is fieldwork, not questionnaire design.
When Is a Market Size the Wrong Tool?
Often enough to ask before the work starts.
- When the category is already measured continuously. Circana and Euromonitor track established consumer categories at a granularity no commissioned study will match, and buying that data beats building the number.
- When the decision is about price. A market size tells you how many people could buy, not what they will pay. Substituting one for the other is how launch prices get set from spreadsheets.
A market size estimates an opportunity. It does not forecast a result. Care makes it more defensible, never a plan.