Seven Korean Provinces Lost Ground on Internet Use

95.0% nationally — and seven provinces going the other way

Korea’s national internet use rate has climbed steadily: 94.0% in 2023, 94.5% in 2024, 95.0% in 2025. Half a point a year, three years running. It is about as tidy a national trend line as official statistics produce.

Underneath it, 7 of the 17 provinces finished 2025 below where they started in 2023. Daegu fell 2.8 points, Ulsan 2.8, Gyeongnam 2.0, Gyeongbuk 1.8, Chungnam 1.7, and Daejeon and Chungbuk 0.6 each. Daegu is the striking one: at 98.4% it was the most connected place in Korea in 2023, and by 2025 it had dropped to 95.6% and fifth position.

Meanwhile Gyeonggi gained 3.7 points and Gwangju finished 2025 as the country’s most connected region at 97.8%. The national line went up because the gains were bigger than the losses, not because everywhere improved.

Internet use by province, 2023–2025

This is a sample survey, and it wobbles

Before drawing conclusions from any single province’s movement, it is worth looking at how the series behaves year to year. Gangwon reads 90.9%, then 88.9%, then 91.9% — down two points, then up three. No plausible real-world change in internet access moves that way in twenty-four months.

That is a sampling artefact, and it sets the scale for everything else in the table. This survey interviews a sample of households, not the whole population, so each published figure carries an error band. Movements of a point or two are inside the noise. Seven provinces are down over three years, but several of them are down by margins the survey cannot resolve.

• Treat ±1 point as indistinguishable from no change. That covers Daejeon (−0.6), Chungbuk (−0.6), Sejong (+0.6), Busan (+0.3), Jeonnam (+0.5), Gangwon (+1.0) and Incheon (+1.0).
• The movements worth taking seriously are the large, consistent ones. Daegu (−2.8) and Ulsan (−2.8) each declined in both consecutive years, which is harder to dismiss than a single-year jump.
• Gyeonggi’s +3.7 is the most solid gain — up in both years, and the largest change in the table in either direction.
• Rank is more fragile than level. Daegu moving from 1st to 5th sounds dramatic, but the four provinces now above it are bunched within 1.6 points of each other; small shifts reshuffle the order easily.

We are deliberately not offering an explanation for why Daegu and Ulsan fell. A facility dataset cannot tell you that, and a survey movement of this size has several possible causes — genuine change, sample composition, or definitional drift among them. The honest statement is that the survey recorded a decline in those two places over three years, and why is a question for the people who run the survey.

The regional gap is narrowing

The more robust story in this table is not any one province, it is the spread. The distance between Korea’s most and least connected province was 8.5 points in 2023, 8.6 in 2024, and 6.7 in 2025.

That compression happened from both ends. The 2023 leader, Daegu at 98.4%, came down; the 2023 laggard, Jeju at 89.9%, went up 2.1 points. Three provinces sat below 91% in 2023 — Jeju, Jeonnam and Gangwon — and all three are above it now. The floor is rising faster than the ceiling, which is exactly what convergence looks like — and it is a more meaningful finding than any individual province’s wobble, because it is visible in the shape of the whole distribution rather than in one estimate.

One thing the table refuses to show is a capital-city premium. Seoul has never led this series. It sat at 93.2% in 2023 — fourth from the bottom of all seventeen provinces — and its climb to 94.7% leaves it ninth in 2025, mid-table, below Gwangju, Gyeonggi, Daejeon, Sejong, Daegu, Busan, Jeonbuk and Chungbuk. Internet use is close enough to universal across Korea that the usual capital-versus-country story simply does not appear in the numbers.

It is also worth keeping the scale in view. The gap between Korea’s best and worst provinces is now under seven points, on a base of more than 91%. Even Jeonnam, in last place, reports that more than nine in ten residents use the internet. Korea’s regional digital divide, measured this way, is a narrow one — which is why the more revealing divides in Korean digital access are by age and by disability rather than by province.

The top and bottom provinces are one row in our own data

There is a wrinkle here that we ran into ourselves, and it is a useful warning for anyone joining Korean datasets together.

Gwangju is first in this table at 97.8%. Jeonnam is last at 91.1%. They are neighbours — a metropolitan city and the province surrounding it — and in our own facility data they are merged into a single region. Every count we publish for “Jeonnam & Gwangju” spans two places that sit at opposite ends of the national internet-use table, 6.7 points apart.

• We cannot un-merge them. Splitting a rate requires the population weights behind each half, which this table does not publish. Averaging 97.8% and 91.1% would produce a number that describes nobody.
• So we do not join these two datasets. This survey is reported at its own 17-province granularity and our facility counts at our 16-region one, and we keep them in separate tables rather than forcing a match.
• The general rule: join on names, never on region codes. Korean statistical files disagree about province codes often enough that a code-based join produces wrong answers without producing an error — and a merged region like this one produces wrong answers even when the names match.

This is the kind of thing that is invisible until you try to combine sources. A national figure and a provincial figure can both be correct and still refuse to line up, because the two publishers drew their boundaries differently. Noticing that before publishing is most of the work.

Sources and method

• Source: the 인터넷이용실태조사 (Survey on Internet Use), conducted for the Ministry of Science and ICT by the National Information Society Agency. Provincial rates for 2023–2025 are taken from the survey’s statistical tables on KOSIS, Korea’s national statistics portal; the survey’s own statistical data files are published at 한국지능정보사회진흥원_인터넷이용실태조사 통계자료.
• The published table mixes 17 individual provinces with three regional subtotals (Chungcheong, Jeolla, Gyeongsang). We use the 17 provinces only; including the subtotals would double-count six of them.
• Figures are shares of the surveyed population using the internet, as published, to one decimal place. Percentage-point changes are computed from those published values.
• This is a sample survey. Year-to-year movements of one to two points should be read as within sampling variation rather than as measured change.