- revIQ, powered by Groundflow, reads 30 million fiscal receipts a month, about 5% of everything issued in São Paulo State.
- Food retail alone accounts for 10.6 million of those monthly receipts, from 8.3 thousand observed stores.
- To tell whether a price went up or down, 260 municipalities and 1,888 neighborhoods in the state already have enough monthly reads.
How much of the state fits in a receipt
Every time someone checks out at a supermarket in São Paulo, the register produces a fiscal receipt recording what was bought, what it cost, which store sold it and when. The state issues around 20 million of these receipts a day, close to 7 billion a year. Anyone who can read a slice of that flow gets a picture of consumption built from purchases that actually happened, with no reliance on a survey respondent’s memory or goodwill.
revIQ, powered by Groundflow, currently reads 30 million receipts a month, roughly 5% of everything the state issues. At first glance that looks like a thin slice, and the question behind this study grows out of that impression. How much of São Paulo can you actually see through 5% of its receipts? In which municipalities and neighborhoods is the volume already large enough to support a publishable number, and where does it still fall short?
Answering it requires agreeing first on what a reliable number is. Statistics settles that with two measures. Confidence tells you how often an estimate is right, and error tells you how far it can stray from the true value. A result at 90% confidence with a 5-point error lands within that range nine times out of ten. Since every use of data calls for a different level of rigor, the study doesn’t impose a single precision. You choose yours, and the simulator shows how far the data reaches with it.
Read the full study.
Fill this in once and every study in the Índice HYPR opens for you in this browser. No password, no account.
- How we measure coverage
- A market of 7 billion receipts a year
- Coverage simulator by precision
How we measure coverage
Before any math, the coverage map rests on four methodological choices. They define the universe of stores, which receipts count, the point at which an area becomes publishable, and how major chains are handled relative to independent retail.
- 01Universe
Receita Federal's registry sets the denominator.
We count active establishments in the state, from Brazil's Federal Revenue Service registry, and group them by primary activity code (CNAE). Food covers codes 4711, 4712 and 4729-6/02; pharmacy, 4771; restaurants, 5611; bakeries and dairy, 4721. The remaining codes in division 47 make up other retail.
- 02Reads
Only receipts actually read are counted.
We use receipts issued from May to July 2026, the three most recent months with complete reads, totaled month by month in every cell of the map: the state, each municipality and each neighborhood.
- 03Threshold
A cell becomes publishable once it reaches the minimum sample.
The calculation uses the chosen confidence and error and assumes the worst case, a 50% proportion. Presence takes a single receipt, while trend, average ticket, share and benchmark call for progressively larger samples.
- 04Segment
Major chains and the long tail are measured separately.
A major chain is any root CNPJ (the company-level tax ID) with 20 or more stores in the state. Because the two groups behave like distinct markets, each gets its own coverage count.
The worst case deserves a word of explanation. The margin of error on a proportion peaks when that proportion sits at 50%, where the result is equally likely to tip either way. Sizing the minimum sample for that scenario guarantees the promised margin holds for whatever proportion the data turns up later.
A market of 7 billion receipts a year
Before looking at the map, it helps to size the market. Sefaz-SP, the São Paulo State Treasury, has reported receipt issuance figures through 2020; for the years after that, we project growth of 2.5% a year. The result is a huge, slow-growing market that has already passed 7 billion receipts a year.
- Sefaz-SP (reported)
- Estimate
Source: Sefaz-SP (Nota Fiscal Paulista), 2014–2020; estimated at 2.5% a year from 2021.
revIQ’s reads move at a different pace. The 2025 and 2026 volumes are actuals, and they show the base jumping from 91 million to 230 million receipts a year. The estimate for 2027 is 360 million, and from there the projection assumes 5% annual growth, twice the market’s pace. In practice, the share of the state the data can see should keep widening every year.
- Actual
- Estimate
Actual for 2025 and 2026 (revIQ, powered by Groundflow); estimated from 2027, at 360 million and 5% annual growth.
Reading 5% of a market this size can sound modest. In statistics, though, what supports an estimate is the number of observations inside each cut, far more than the fraction of the whole that cut represents. With millions of receipts a month, there is plenty of data to answer a great many local questions, and coverage comes to depend mostly on how much rigor each question demands.
Reading 5% of what the state issues is already enough for hundreds of municipalities and thousands of neighborhoods, as long as the question fits the precision you choose.
Coverage simulator by precision
The simulator starts from five use cases, ordered from the simplest to the most demanding. Knowing whether a store is open and selling takes just one receipt in the period. Telling whether something went up or down from one period to the next needs 90% confidence and a 5-point error. Average ticket and category mix raise confidence to 95%, and brand share, the market-research standard, tightens the error to 3 points. At the top of the scale sits the institutional benchmark, meant for press-ready and board-level numbers, which requires 99% confidence and a 2-point error. If you prefer, set confidence and error by hand.
Then pick the geography, frequency, sector and segment, and the panel shows how many municipalities and neighborhoods already have enough reads under those conditions. The calculations run on the actual distribution of receipts in each cell. A statewide average would hide how unevenly the data spreads across the map, and that unevenness is precisely what decides where a number can be published.
Loading the simulator…
Major chain = root CNPJ (company-level tax ID) with 20 or more stores in São Paulo State. Source: revIQ, powered by Groundflow, and Receita Federal (active establishments); receipts issued May to July 2026.
In practice, the same base of reads can answer very different kinds of questions. Tracking the direction of prices and sales is already possible, month by month, in 260 municipalities and 1,888 neighborhoods, while benchmark-grade numbers remain limited, for now, to the busiest areas. Because reads are growing faster than the market, that frontier should keep moving toward smaller cities and neighborhoods.