74% of Older Koreans Can Use a Map App

The assumption underneath every map app

Every “find your nearest toilet” service — ours included — rests on an unstated premise: that the person who needs one can open a map on a phone and read it. It is worth knowing how often that premise is false, and Korea happens to measure it directly.

The Digital Information Divide Survey, run annually by the National Information Society Agency (NIA) for the Ministry of Science and ICT, asks respondents whether they can use navigation, online map services and transit information. Answers run on a four-point scale from “not at all” to “very much so”. It is a national approved statistic, and the underlying respondent-level data is public.

We aggregated the raw responses for 2025 rather than using a published index, because the published figure is a composite score and we wanted this one question on its own. Here is what it says.

One limitation to hold onto throughout: this is self-assessment, not a test. Respondents are reporting how capable they feel, and people systematically misjudge that in both directions — some overstate, some are unduly modest, and the tendency differs by age and culture. Read every figure below as “say they can”, not “demonstrably can”.

Who can use an online map, by group

The headline number is the one at the bottom. 26% of older adults and 26.3% of farming and fishing households say they cannot use an online map. For roughly one person in four in those groups, “just look it up on the map” is not advice — it is a dead end.

Two findings that run against the assumption

First: marriage immigrants and North Korean defectors report the highest ability of any group — higher than the general public. 92.5% and 91.2% against 88.6%. If you expected migrants to sit at the bottom of a digital-divide table, this question does not support that. People who have moved countries and have to navigate an unfamiliar one are, unsurprisingly, heavy users of maps.

Second, and this is why the second column exists: on confidence, the ranking flips. Only 32.1% of marriage immigrants pick the strongest answer, against 41.5% of the general public. The honest summary is that they can use maps but do not feel expert at them — which is exactly the profile of someone who would benefit from clearer interfaces and worse served by dense ones.

We report both columns everywhere because reporting one is genuinely misleading. When we first tabulated this by province using only the strongest answer, Seoul’s disabled respondents came out near the bottom at 10.7%. Using both positive answers, Seoul came out top at 94.7%. Same data, opposite conclusion — the confidence distribution varies by region enough to invert the ranking.

Where you live matters as much as which group you are in

The survey also records where respondents live. Taking the disabled respondents — the group with a large enough sample to break down geographically — and splitting them by province produces a spread wider than the gap between any two groups in the table above.

Seoul’s disabled respondents report 94.7%; Gangwon’s report 60.2%. That is a 34.5-point spread within a single group — nearly twice the 18.8-point spread between the highest and lowest groups nationally.

In other words, a disabled person in Seoul is more likely to be able to use a map app than the average member of the general public, while two in five disabled people in Gangwon cannot. Being disabled predicts less about digital access than living somewhere rural does. Any policy or product that treats “people with disabilities” as one audience with one level of digital access is working from a picture the data does not support.

What this means for services built on maps

We build a map app, so this is uncomfortable data to publish, which is a reasonable sign it is worth publishing. A few things follow from it.

• A map is not an accessibility strategy. For a quarter of older Koreans, an app that can only be operated by reading a map is unusable regardless of how good the data behind it is.
• Physical signage still does the heavy lifting for the groups least able to use apps — which is why we care about what is recorded for braille blocks and guide staff, not only what is on the screen.
• Voice and text entry matter more than they look. For someone who can operate a phone but not confidently read a map, being able to say what they need is the difference between the service working and not.
• Large type and high contrast are not a nice-to-have. The groups with the lowest confidence scores overlap heavily with the groups most affected by presbyopia and low vision.

It also means the honest description of a service like ours is narrower than the marketing version. We can help the roughly nine in ten people who can already work a map find things faster. We cannot, on our own, reach the one in four older adults who cannot — and pretending otherwise would be the easiest thing in the world to write and the least useful.

The people that gap describes are, awkwardly, the same people most likely to need a public toilet urgently, least able to walk an extra ten minutes to the wrong one, and most likely to live where the alternatives are thinnest. The population an app like this serves worst overlaps heavily with the population that needs it most. We do not have a fix for that. We would rather say it than let a chart imply otherwise.

Sources and method

• Source: 2025 Digital Information Divide Survey, Ministry of Science and ICT / National Information Society Agency (NIA). National approved statistic no. 120017. Respondent-level data published as 한국지능정보사회진흥원_디지털정보격차실태조사 통계자료 on Korea’s public data portal; survey overview at NIA.
• Method: we aggregated the raw responses to the survey’s navigation-and-online-maps question, applying the survey’s own weights. These are our own calculations from the microdata, not a published NIA index.
• Weighted and unweighted results differ by under 0.5 percentage points in almost every cell; the exception is the confidence figure for people with disabilities (18.5% weighted against 22.2% unweighted).
• We do not report cells with fewer than 100 respondents.