The India Data Atlas
Every guesstimate is a chain of multiplications hanging off one or two anchor numbers. Most candidates memorise a list of them and get caught out by the same three things: the anchor has drifted, they multiplied a per-household rate by the population, or they could not say where the number came from. So this sheet does the opposite of a list. It is built as twelve slides, each one chart-led with a single takeaway at the foot; every figure carries where it came from and how old it is; and there is a search box, because seventy-seven numbers is well past the point where scrolling helps.
TL;DR · Key Takeaways
The spine, in one breath
- 1.48 bn people · ~320 mn households · ~645 mn workers · GDP ₹346 lakh crore. Almost everything funnels from one of these four.
- 37% urban. Median age about 29. Fertility 1.93 — already below replacement. Young is a stock, not a flow.
- Average spend is ₹4,122 rural and ₹6,996 urban per person per month. Not ₹19,500 — that is output per head, and it is not in anyone's pocket.
- Agriculture: 43% of the jobs, 18% of the output. Output share and employment share are different questions.
- Connections are not people. 1,289 mn mobile connections, 1,194 mn active, ~680 mn smartphone users, urban tele-density 152%.
- Annual sales × life ≈ stock. It closes for two-wheelers and cars to within a few per cent. Learn the move, not the numbers.
- Round in powers of ten, name every assumption, and always build the answer a second way.
How to use it in thirty seconds
Pick the anchor, not the answer
People, households, workers, or GDP. Say which one out loud and why. Almost every question funnels from one of the six in Start here.
Cut before you multiply
Age, urban or rural, income. An urban rate applied to 1.48 bn people is the single most common way a good structure produces a wrong answer.
Round hard, in powers of ten
Work in crores and exponents. ₹1 lakh crore is 10¹². Nobody has ever lost a case for rounding; plenty have for a misplaced zero.
Close the loop
Rebuild the answer from a different anchor and say whether the two agree. If they land within about 2×, say so — that sentence is worth more than the estimate itself.
The atlas
77 numbers, every one sourced · September 2026
Start here
The six anchors
Nine out of ten guesstimates hang off one of these. Get the anchor right, then round everything downstream without guilt.
Population
Official estimate1.48 bn
≈148 crore
2026 · UN World Population Prospects 2024 revision
Households
Our arithmetic~320 mn
≈32 crore
2026 (derived from Census 2011) · Derived: Census of India 2011 household count × UN WPP 2026 population
Urban share
Official estimate36.9%
≈545 mn urban · ≈930 mn rural
2024 · World Bank
People who work
Our arithmetic~645 mn
≈64 crore
Jan–Dec 2025 · Derived: PLFS 2025 worker-population ratio × UN WPP age structure
GDP (nominal)
Official estimate₹346 lakh crore
≈ $4.15 trillion
FY2025-26 (Provisional Estimates) · MoSPI
GDP per person
Our arithmetic₹2.35 lakh / yr
≈₹19,500 a month · ≈$2,810
FY2025-26 · Derived: MoSPI GDP ÷ UN WPP population
Everything funnels from the top
Four cuts of the same population. Pick the one your question actually needs — using "1.48 bn" where you meant "644 mn" is a 2.3× error before you have multiplied anything.
So whatAn interviewer forgives a 20% rounding error all day. A 10× anchor error ends the case in the first minute.
People
The shape of 1.48 billion
Age decides who can buy what. Cut the population before you multiply by it.
If India were 100 people
Fertility rate
1.93
The rule marks replacement level. India is already under it — the cohorts behind today’s young population are smaller.
Cohort flow each year
Net population growth
29
median age, years
72.7
life expectancy, years
80.9%
literacy, age 7+
940
women per 1,000 men
So whatIndia's fertility rate is 1.93 — already below the 2.1 replacement line. The young country is a stock, not a flow: the cohorts coming behind are smaller.
Money
What a household actually spends
The most expensive assumption in any guesstimate is that the average Indian is middle class.
Rural spend per person
Counted₹4,122 / month
₹4,247 including free welfare items
2023-24 · MoSPI
Urban spend per person
Counted₹6,996 / month
₹7,078 including free welfare items
2023-24 · MoSPI
Output per head against what people actually spend
Both are rupees per person per month. GDP per capita is what the economy produces per head; MPCE is what a surveyed household actually spends. Quoting the first as though it were the second is the most expensive slip on this page.
Rural against urban
Per person
urban 70% higher
Per household (×4.5)
the whole family budget
The household figures are ours: MPCE × an assumed household size of 4.5. Say the assumption aloud and nobody argues.
47%
food share of rural spend
40%
food share of urban spend
61.5%
private consumption, share of GDP
30%
investment, share of GDP
So whatThe average Indian household spends about ₹18,500 a month in a village and ₹31,000 in a city — everything, for everyone in it. Price against that, not against your own bill.
Work
Who actually earns
Working-age is not working. Confusing the two inflates consumer numbers by about 60%.
Where the jobs are, and where the output is
Agriculture & allied
−25.1 pts
Industry
+3.0 pts
Services
+22.0 pts
Job shares are PLFS 2025 (industry = manufacturing + construction; services = the remainder). Output shares are national accounts at current prices.
59.3%
in the labour force, 15+
57.4%
actually working, 15+
3.1%
unemployed · urban women 6.4%
45.9 / 27.6%
female participation, rural / urban
1 more number on this slide
24.8 crore students
The school system
So whatAgriculture is 43% of the jobs and 18% of the output. That one sentence is worth more in a group discussion than any other statistic on this page.
Economy
What the economy is made of
Sector shares give you a top-down route into almost any industry question.
Gross value added by sector
At current prices, 2024-25. At constant prices agriculture is smaller again, nearer 14%.
7.7%
real GDP growth, FY2025-26
8.9%
nominal GDP growth
357.7 mt
foodgrain production
239.3 mt
milk production
142.8 bn
eggs produced
2022-23
GDP base year — rebased in 2026
So whatServices are 55% of output. Industry is 27%. Agriculture is 18% — and employs more than twice its output share. Output and jobs are different questions.
Digital
Connections are not people
The most over-quoted numbers in Indian guesstimates. Almost all of them count SIMs, not humans.
Mobile connections
Counted1,289 mn
≈129 crore
April 2026 · TRAI
Broadband connections
Counted1,073 mn
wireless 1,026.6 mn · wired 46.8 mn
April 2026 · TRAI
UPI transactions a day
Counted791 mn
≈79 crore a day
August 2026 · NPCI
From SIMs down to humans
The top bar is the number everyone quotes. The bottom one is the number to use.
Tele-density
Vertical rule: 100% — one connection per person
Urban India is past one connection per person. Any model that treats connections as users is 30–50% too high there.
791 mn
UPI transactions a day, Aug 2026
₹1,217
average UPI ticket
₹314 lakh cr
UPI value, FY2025-26
152 mn
smartphones shipped a year
So whatUrban tele-density is 152%. There are more mobile connections in Indian cities than there are urban Indians — so any estimate built on "connections = users" is already 30–50% too high.
Mobility
Stock and flow on the road
Vehicles are the cleanest place to learn the stock-and-flow move, because both numbers are published.
What India actually drives
Roughly five two-wheelers for every car — about 185 against 35 per 1,000 people. More than 60% of households own a two-wheeler; under 10% own a car.
Stock from flow — the move worth learning
Two-wheelers on the road
Cars on the road
2.83 cr
vehicles sold, FY2025-26
2.0 cr
railway passengers a day
1,00,266
petrol pumps
93.9 / 42.7 mt
diesel / petrol a year
So whatAnnual sales × vehicle life ≈ the vehicles on the road. Two-wheelers: 2.17 crore a year × 12 years ≈ 260 mn — which is exactly the registered stock. When a chain closes like that, you can say so out loud, and the room notices.
Per person
Per person, per day
These are the multipliers. The discipline is choosing the right population to multiply — national, urban, or one city.
Per person — each on its own scale
135 L/day
water — urban design norm (metros 150–200)
1,395 kWh/yr
electricity, ≈3.8 a day
471 g/day
milk availability
242 kg/yr
foodgrain, as produced
0.35 kg/day
municipal waste, urban average
92 kg/yr
road fuel
103/yr
eggs
25–30 GB/mo
mobile data per user
Deliberately tiles, not one bar chart: 135 litres and 3.8 kWh do not share an axis, and a chart that pretends they do is lying about the comparison. The discipline that matters here is picking the right population to multiply by — national, urban, or one named city.
So whatAlmost every wrong guesstimate applies an urban rate to the whole country. Water at 135 litres a day is a design norm for a piped city, not what 1.48 bn people receive.
How many
How many of everything there are
When the question is "how many X in India", one of these is usually the denominator you need.
One city, three populations
Mumbai, Census 2011. The widest and narrowest definitions differ by 1.9× — enough to swing a city guesstimate on its own. Name the one you are using before you multiply.
Counts worth knowing cold
640
districts in 2011 — nearer 800 today
6.4 lakh
villages
7,935
towns
248.8 mn
households, last hard count
14.72 lakh
schools · 24.8 cr students
≈13 mn
kirana stores — estimate, never counted
1 more number on this slide
Delhi NCR ≈58 mn · Mumbai ≈24 mn
The biggest cities
So whatA city has three different populations depending on whether you mean the municipal corporation, the urban agglomeration or the metropolitan region — and they can differ by two times. Say which one you are using before you multiply.
Arithmetic
Lakh, crore, and time
Mis-converting crore is the quietest way to be wrong by a factor of ten.
The zero ladder — where guesstimates quietly die
10⁵
1 lakh
100,000 · 0.1 million
10⁷
1 crore
10,000,000 · 10 million
10⁹
100 crore
1,000,000,000 · 1 billion
10¹²
₹1 lakh crore
₹1 trillion ≈ $12 bn
Work in exponents, not words. Writing the power of ten down is the single cheapest error-check available to you.
Time constants
1,440
minutes a day · ≈1,000 waking
8,760
hours a year · round to 8,000
250 / 300
working days — five-day / six-day
4.3
weeks a month, not 4
So whatWork in powers of ten, not in words. 1 crore is 10⁷. ₹1 lakh crore is 10¹² — a trillion rupees, roughly $12 bn. Write the exponent down and the conversion stops being a risk.
Defaults
When you have nothing to go on
State one of these out loud, then proceed. A named assumption is a strength; a silent one is a trap.
The defaults, and what they are worth
4.5
household size · urban 4.2, rural 4.8
76%
of people are 15 or over
≈680 mn
smartphone users — not a billion
3.5–4.5 yrs
phone replacement cycle
12 / 11 yrs
two-wheeler / car life
8–10 yrs
TV and fridge replacement
4–6/week
paid meals out, urban
≈25
pupils per teacher
None of these are measurements. They are the numbers that make other numbers reconcile — the vehicle lives, for instance, are simply what makes annual sales agree with registered stock.
So whatInterviewers are not testing whether you know the number. They are testing whether you notice you needed one — and whether you said so.
Cross-checks
Close the loop
The move that separates a good answer from a memorised one: build the number twice, from two unrelated directions, and say whether they agree.
Build it twice, from unrelated directions
Two-wheelers on Indian roads
Cars on Indian roads
People who actually use a smartphone
What Indian households spend in a year
UPI transactions per adult per day
Three of these close. Two do not — and saying why is worth more than the estimate itself.
3 more numbers on this slide
Surveys capture ≈43%
Surveys against national accounts
Registered overstates
Registered against running
Last full count: 2011
The census gap
So whatYou do not need to be right. You need to be able to show you would have noticed if you were wrong.
Practise reaching for the right one
Which anchor, and what would you say out loud?
Do not compute anything. For each prompt, name the anchor you would start from, the one cut you would make first, and the assumption you would state aloud. Then check yourself.
How many refrigerators are sold in India in a year?
What is the annual revenue of all petrol pumps in India?
How many people in India actually use a smartphone?
Size the market for packaged milk in a tier-1 city.
How much municipal waste does a metro generate in a day?
How many UPI transactions does one person do in a day?
Read the vintage before you quote the number
India's last full census was 2011. Every household, village, town and city figure here is either fifteen years old or estimated forward from something that is; the 2027 census will move several of them. Three more things to keep straight: (a) GDP was rebased to 2022-23 in 2026, so pre-2026 write-ups are on a different series — do not mix them in one argument; (b) consumption surveys and national accounts disagree by roughly 2×, by design, so name which basis you are on; (c) registered vehicles overstate vehicles running, because scrapped ones are rarely de-registered. Saying any of these in a room marks you out faster than a correct number does.
Going deeper → the Industry Primers
These are national anchors for fast sizing, not sector depth. When a guesstimate is really about one industry, open its Industry Primer for ARPU, GMV, penetration, capacity and sales-mix rather than reinventing them here: Telecom, Payments, E-Commerce, Food Delivery, Automobile, Electric Vehicles, FMCG, Retail, Banking, Healthcare, Power, Oil & Gas — plus around fifteen more under Industry Primers in the sidebar.