87% of People With Worries Get No Support. What to Know Before You Build a Service for Loneliness

Japan has a minister in charge of loneliness and isolation. It has had one since February 2021. On 1 April 2024 the Act on the Promotion of Measures to Address Loneliness and Isolation came into force, and now local governments are expected to act on it too, not only the national government.

Since then I hear about more and more services aimed at loneliness. Watching over older people, helplines for young people, mental health support for employees. They come in every shape.

I build an AI companion app for iPhone myself. It is called AnimaEcho. You talk, and an animated character answers in her own voice, in real time. Building it meant reading a lot of the data in this area.

There are a few numbers I would want anyone to know before they start planning.

About 4 in 10 feel lonely

The Cabinet Office published the 2025 edition of its national survey on social connection in April 2026. It sampled 20,000 people aged 16 and over at random across Japan and got 11,873 valid responses.

4.5% said they feel lonely often or always. 13.7% said sometimes and 19.5% occasionally. Together that is 37.7%, or about 4 in 10.

By age, the share who feel lonely often or always is highest among people in their thirties, at 6.1%. People in their twenties to fifties sit between 5.5% and 5.7%. People in their seventies are at 2.5%.

That is a little different from the picture of loneliness as a problem of old age. The people having the hardest time are in the middle of their working lives.

87% get no support

For anyone planning a service, I think this is the number that matters most.

The survey asks people who feel anxiety or worry in daily life whether they receive support from government bodies, NPOs or similar. 87.0% said no. That is up from 85.1% the year before.

The most common reason, at 63.1%, was that they do not need support.

That is the whole difficulty of this field in two numbers. People have worries, but they do not see themselves as people who need help. People feel lonely, but they will almost never walk up to a counter labeled "for lonely people."

So the moment you call it a loneliness service, the people you most want to reach walk away.

A different way in

The same survey also has a hint.

It compares loneliness with how often people eat with someone else. Among people who eat with someone almost every day, 2.7% feel lonely often or always. Among people who almost never do, it is 17.3%. More than six times higher.

9.7% of people never meet family or friends they do not live with in person.

Loneliness rarely shows up as a special problem. It shows up in ordinary days. Eating dinner alone. Nobody to tell what happened today.

So the way in works better when it looks like ordinary life. Something shaped like having dinner or telling someone about your day reaches further than something shaped like asking for advice. If a service looks like something people already want to do, they can use it without putting a label on themselves.

What I decided for AnimaEcho

A few decisions I made while building AnimaEcho. They may help if you are planning a service in this area.

Keep the word loneliness out of the front. AnimaEcho is described as someone who listens and answers in her own voice. The reason to open it is that you want to talk about your day. That is enough.

Talk by voice. Typing is work. More evenings than you would think go by in a room where someone never says a word out loud. Talking and hearing a voice back means something.

No feed. There is nothing to scroll. You open it, you talk, you close it.

A daily free allowance. Free talk time resets every day. I wanted it to be a place you drop in on and then leave, rather than a place you stay.

Do not measure time spent

The most important decision is how you measure success.

Most apps are judged on time spent and number of opens. Do that with a service for loneliness and you build a product that depends on the loneliness of the people using it.

Someone talks for ten minutes, feels a little lighter and puts the phone down. Maybe they message a friend afterwards, maybe they just sleep better. That is a good outcome. If the app has become the only one they talk to all day, it has failed.

One more thing, stated plainly. An AI companion is not a friend and not a counselor. It does not replace human connection, and it should not pretend to. If you catch yourself writing "solves loneliness" in a proposal, stop and think again.

AnimaEcho is free on the App Store. There is more at animaecho.com.

If you are planning a service around loneliness or watching over people and want to talk it through at the design stage, get in touch.

Sources

• Cabinet Office, Office for Policy on Loneliness and Isolation, National survey on loneliness and isolation (2025 survey on social connection). 20,000 people aged 16 and over sampled at random, 11,873 valid responses, survey date 1 December 2025, published 14 April 2026. Figures from the summary of results, figures 1, 3, 5, 7 and 9.
• Cabinet Office, Act on the Promotion of Measures to Address Loneliness and Isolation. Enacted 31 May 2023, in force 1 April 2024.