Live Selling Playbook
Create account

BREAKDOWNS · Agency PLAYBOOK

Xinba's comeback broadcast: the anatomy of a redemption stream

Updated 2026-10-16 · Sources & methodology noted inline

THE CREDENTIALS (public record): After the bird's-nest scandal (see our companion breakdown), Xinba served his 60-day Kuaishou ban, stayed off camera for roughly 100 days total, and returned in spring 2021 with one of the most-watched single streams in Kuaishou history — third-party trackers and business media reported ¥2.043 billion in sales for the comeback session. The build-up was a campaign in itself: a public letter, city-wide outdoor advertising bought by his team, a red-carpet return in which he and his apprentice-hosts bowed to the camera ('welcoming every user home'), and a stream engineered as ceremony rather than sale. Whatever you think of the operator, the redemption stream is a documented case study in rebuilding a burned trust account in a single broadcast — this breakdown deconstructs it as mechanics.

WHAT HE DID (the sequence): the comeback ran in deliberate phases. Weeks before: silence — no casual posting, no teasing, so the return registered as an event. Days before: the apology infrastructure — the bow, the letter, the advertising — all framed as welcome-home, not please-forgive-me. The subtlety matters: 'welcoming users home' positions the audience as the ones who left and chose to return, which restores their agency and pride; 'forgive me' positions them as judges, which reminds them of the crime. During the stream: massive giveaway mechanics (reported ¥10M+ in free products across the session), apprentice hosts carrying product segments while Xinba opened and closed, and a product slate heavy on low-claim, high-value staples rather than the health-adjacent categories that caused the scandal.

This is a Agency playbook

67% of this playbook is locked. Unlock it — and the rest of the breakdowns library — with a membership.

Sign in to unlock →

Run the numbers: Free profit calculator·Free script generator