Роботы-компаньоны умирают к 30-му дню: что не так с нишей AI-компаньонов

Researcher Jeff Ding, in issue #366 of the ChinAI newsletter, identified a structural problem in the AI companion market: most companion robots "die" by day 30 — they fail to retain users beyond one month.

What "death at day 30" means

Day thirty is the standard benchmark for digital products. The Day 30 Retention metric shows what percentage of users return to an app or device one month after their first launch. According to ChinAI's observations, most companion products fail to clear this benchmark.

"Most companion robots die by day 30," — that is how

Jeff Ding, author of the ChinAI newsletter covering artificial intelligence developments in China, formulated the key thesis.

China's AI companion market is one of the most active segments of consumer AI. It includes both software products (chatbots with "personality," voice assistants with emotional interfaces) and physical devices — small robots for home and office use. The number of new launches in this niche keeps growing, yet the retention problem remains systemic.

Why the first month turns out to be fatal

AI companions are built on an emotional hook: the promise of a living conversational partner, personalization, and attachment. This is what sets them apart from utilitarian assistants — and it is precisely here that the main trap lies. Emotional attachment forms slowly, while the "novelty effect" fades within the first two to three weeks.

By day 30, most users have already encountered repetitive responses, failures in conversational context, and predictable reactions. Disappointment arrives before attachment has a chance to form — and the user leaves.

For physical companion robots, the disappointment threshold is further lowered by high prices: the device creates inflated expectations, and when the "magic" of the first month fades, it becomes extremely difficult to justify continued use.

Why ChinAI writes about this

The ChinAI newsletter, which Jeff Ding has been running for several years, specializes in content that falls outside Western headlines: structural market patterns, trend analytics, breakdowns of failures rather than just successes. Issue #366 fits this logic: the companion niche's problem has long been obvious to insiders, but is rarely discussed publicly.

Similar dynamics can be observed outside China as well — Western startups like Replika and Character.ai also see sharp engagement spikes followed by churn. However, competition in the niche is especially fierce in China, and the sheer number of simultaneously launched products makes the retention problem most visible there.

What this means

Failing the D30 metric makes most business models in the AI companion niche structurally vulnerable: without returning users, neither subscription monetization nor organic growth through referrals works. Until companies learn to build long-term emotional value or radically reduce technical limitations, the cycle of "launch — hype — failure" will repeat itself again and again.