Half the Country Offline: The Subscriptions Are There, the Users Are Not

Executive Summary. In 2024, 53.42% of Bangladesh's population used the internet, the lowest of six Asian economies and 19.36 percentage points below the peer median of 72.78%. Against a population of 173.6 million, that leaves 80.8 million people offline. The standard explanations do not survive the comparison. Electricity is not the binding constraint: 99.5% of the population had access in 2023, level with the peer median. Nor is the physical network: Bangladesh reported 8.09 fixed broadband subscriptions per 100 people, third of the six and 1.64 times the peer median, ahead of Indonesia (4.92), India (3.15) and Pakistan (1.47). It also reported 108.1 mobile subscriptions per 100 people, ahead of India's 79.35 and Pakistan's 76.85. Indonesia, India and Pakistan all put a larger share of their people online. That is the finding: on both counts of internet-capable connections, Bangladesh places in the middle of the group; on the count of people connected, it places bottom. The gap is not in the network. It is between a subscription and a user, and none of the World Bank series used here can tell you what sits inside it.

Eighty million people, and the grid reached all of them

In 2024, 53.42% of Bangladeshis used the internet. The comparison group: Thailand 90.87%, Vietnam 84.15%, Indonesia 72.78%, India 64.94%, Pakistan 57.25%. Bangladesh is last of the six, at 73.4% of the peer median of 72.78%, a shortfall of 19.36 percentage points.

The share is the headline; the count is the policy object. Applied to a population of 173,562,364, it leaves 80,842,376 people who did not use the internet in 2024, inside an economy of US$450 billion that grew 4.22% that year. That count is the number of people for whom a service delivered only online is a service not delivered.

The first explanation to discard is power. Access to electricity reached 99.5% of the population in 2023, identical to the peer median of 99.5% and third of the six, with Thailand at 100%, Vietnam 99.8%, India 99.5%, Indonesia 99.4% and Pakistan 95.6%. Electrification is one of the genuine infrastructure achievements of recent decades in Bangladesh, and as a national coverage problem it is finished. Whatever keeps 80.8 million people offline, it is not the absence of a socket.

Source: World Bank World Development Indicators, individuals using the internet (% of population, IT.NET.USER.ZS), 2024.

More broadband than India, less internet than Pakistan

The second explanation to discard is the physical network, and this is where the data stops behaving.

Fixed broadband subscriptions in Bangladesh ran at 8.09 per 100 people in 2024, third of the six and 1.64 times the peer median of 4.92, a margin of 3.17 subscriptions per 100 people above it. Vietnam (23.71) and Thailand (14.89) are ahead of Bangladesh. Indonesia (4.92), India (3.15) and Pakistan (1.47) are behind it. Bangladesh therefore has more fixed broadband per head than three countries that put a larger share of their people online.

Rank the six on broadband density, then rank them on internet use, and five of them come out in nearly the same order twice. Vietnam and Thailand occupy the top two places on both measures (they swap between them), and Indonesia, India and Pakistan follow in identical order on both. Bangladesh is the country that breaks the pattern. It places third on subscriptions and sixth on use.

Source: World Bank WDI, fixed broadband subscriptions (per 100 people, IT.NET.BBND.P2), 2024.

None of this was inherited. Bangladesh skipped the copper era: fixed telephone lines numbered 0.164 per 100 people in 2024, the lowest of the six and about 6% of the peer median of 2.62, against Thailand's 5.45, India's 2.70 and Vietnam's 2.62. Whatever fixed broadband exists in the country was laid as new plant, not upgraded out of an existing telephone estate. The 8.09 figure is a build, and it is the part of the digital programme that worked.

One qualification, raised here rather than left for a critic. A fixed broadband subscription is an operator-reported line, not a person. Household size, sharing arrangements and each country's rules for counting shared or community connections all sit between this series and the user series, and none of them are in the indicator set. What the comparison establishes is that Bangladesh is not short of subscriptions relative to this group. It does not establish that those subscriptions reach anyone.

Subscriptions are not people

Mobile is where the argument would normally be settled, because in this region the internet mostly arrives on a handset. It does not settle it either.

Bangladesh reported 108.1 mobile subscriptions per 100 people in 2024, more than one for every person in the country, fourth of the six and 14.44 below the peer median of 122.5. Thailand (160.6), Vietnam (127.6) and Indonesia (122.5) sit above it. India (79.35) and Pakistan (76.85) sit well below it, and both have higher internet use than Bangladesh does. A country with fewer SIM cards per person than Bangladesh puts more of its people online.

So on both counts of internet-capable connections, Bangladesh places third and fourth of six. On the count of people who used the internet, it places sixth. The residual between those two facts is the entire argument, and the World Bank series used here do not measure it.

Name the candidates for what occupies that residual, and label them honestly. The price of a data package against a day's earnings, the price of a handset able to use it, the ability to read the language an interface is written in, and whether a household's one connected phone is available to everyone in that household are all plausible occupants. Not one of them appears in the indicators behind this essay. A subscription count above 100 per 100 people is also consistent with people holding several SIM cards, which raises the connection count without adding a user. Anyone who tells you which of these binds in Bangladesh is not reading it off this data.

Twenty years of vertical growth that did not close the gap

Bangladesh's own trajectory is not the thing being criticised. In 2005, 0.24% of the population used the internet. By 2024 it was 53.42%: a rise of 53.18 percentage points, and 221 times the 2005 level.

Resist the multiple. A 221-fold increase off a base of roughly a quarter of one percent is arithmetic, not achievement. Any country starting near zero produces a spectacular ratio, and the ratio says nothing about where the country arrived. The level says that, and the level is last of the six, 19.36 points under the median.

Fixed broadband tells the same story about growth and a different one about position: from 0.27 per 100 people in 2010 to 8.09 in 2024, a gain of 7.82 and a multiple of 29.7. That build-out happened, and it shows in the ranking. It did not produce users at the same rate.

Source: World Bank WDI, individuals using the internet (% of population, IT.NET.USER.ZS), Bangladesh against comparators, 2000-2024.

The chart carries the shape of the problem. Bangladesh's curve climbs steeply from the mid-2000s, and across the period plotted it does not converge on the comparator group.

The supply side, as far as this data goes

The production side of the same story has one indicator here, and it is old. ICT goods were 0.05% of Bangladesh's merchandise exports in 2015, the lowest of the six that year, against a peer median of 3.43%, with Vietnam at 29.01% and Thailand at 16.45%. Bangladesh's share was under 1.5% of the peer median.

Two limits before that is used for anything. 2015 is the latest year Bangladesh reports in this series, so it is a decade-old reading and the export mix has moved since. And the indicator covers goods: software and IT-enabled services exports, which is where the country's digital export story is usually told, are not in it. What the series supports is narrow. A decade ago Bangladesh was not exporting hardware, while two comparators had built a large part of their export base on it.

What would change this conclusion

A specific, checkable test. If, by the 2030 WDI vintage, Bangladesh's internet use has passed the current peer median of 72.78% while fixed broadband density stays roughly where it is, the argument here is wrong: the subscriptions were reaching people on a lag, and 2024 caught the country mid-conversion. If internet use in 2030 is still nearer 53.42% than 72.78%, and still below Pakistan's current 57.25%, then a decade of network build-out will have been sold to operators rather than delivered to households.

Three moves, with owners and signals.

1. Measure the offline population before budgeting for it. Connections are administrative data, filed by operators, and they are counted continuously. Use is survey data, and it is the scarcer of the two. Owner: Bangladesh Bureau of Statistics, as a standing module rather than an occasional survey. Success signal: a published internet-use series broken down by sex, division and household income, so that the 80.8 million can be described rather than only counted.
2. Spend the marginal taka on the user side, not another kilometre of backbone. Bangladesh is already third of six on fixed broadband density, at 1.64 times the peer median, while sitting bottom of six on use. That is the signature of a supply-led programme meeting a demand-side constraint. Owner: Bangladesh Telecommunication Regulatory Commission, through licensing conditions and universal-service obligations. Success signal: internet use above Pakistan's 57.25% in the next WDI release without a further jump in broadband density.
3. Do not make a public service online-only while the offline count is this large. Owner: the ICT Division, with the service-owning ministries. Success signal: every service that moves online keeps a funded assisted channel until the internet-use series clears the peer median of 72.78%.

The counterargument

The strongest objection is measurement, and it is a good one. "Individuals using the internet" is not counted the way subscriptions are counted. It comes from household surveys with country-specific questions, recall windows and age cut-offs, and it is estimated rather than observed where survey coverage is thin. On that reading, Bangladesh's 53.42% may reflect a stricter question rather than a smaller internet, and the ordering against Pakistan's 57.25%, the nearest country above it, could reverse under a common definition.

Three answers, one of which is a concession.

First, the gap that matters is not the one to Pakistan. It is 19.36 percentage points to the peer median, and Thailand's 90.87% is not within reach of any plausible definitional adjustment.

Second, the operational claim in this essay is a count, not a rank. 80.8 million people is a large number under any of the definitions on offer, and a public service designed for the connected half of the country misses them whether the true position is sixth of six or fifth.

Third, the concession. If the survey instrument explains part of the distance, then part of the residual this essay locates between subscriptions and users is a statistical artefact, and nothing in this data says how much. That is a real limit. It is also the reason the first recommendation is measurement rather than money.

One further limit applies to everything above. Six countries is a comparison, not a sample, and nothing here identifies a cause. Bangladesh's combination of high subscription counts and low use is consistent with an affordability constraint, with a device constraint, with a literacy constraint, with a measurement difference, and with several other stories this indicator set cannot separate from each other.

---

Data sources: World Bank World Development Indicators, retrieved from the BDPolicyLab data lake, 2026-08-10. Comparators are India, Pakistan, Indonesia, Thailand and Vietnam; "peer median" is the median of the comparators reporting in the stated year, excluding Bangladesh. Years differ by indicator because coverage does: internet use, fixed broadband, mobile subscriptions and fixed telephone lines are 2024, electricity access is 2023, and ICT goods exports are 2015, the latest year Bangladesh reports in that series. All six countries report every indicator used here, so each comparison rests on six observations.

Sources

1. World Bank WDI, Individuals using the Internet (% of population), IT.NET.USER.ZS: https://data.worldbank.org/indicator/IT.NET.USER.ZS
2. World Bank WDI, Fixed broadband subscriptions (per 100 people), IT.NET.BBND.P2: https://data.worldbank.org/indicator/IT.NET.BBND.P2
3. World Bank WDI, Mobile cellular subscriptions (per 100 people), IT.CEL.SETS.P2: https://data.worldbank.org/indicator/IT.CEL.SETS.P2
4. World Bank WDI, Fixed telephone subscriptions (per 100 people), IT.MLT.MAIN.P2: https://data.worldbank.org/indicator/IT.MLT.MAIN.P2
5. World Bank WDI, ICT goods exports (% of total goods exports), TX.VAL.ICTG.ZS.UN: https://data.worldbank.org/indicator/TX.VAL.ICTG.ZS.UN
6. World Bank WDI, Access to electricity (% of population), EG.ELC.ACCS.ZS: https://data.worldbank.org/indicator/EG.ELC.ACCS.ZS