The situation
This customer-service BPO runs at a scale of a thousand-plus seats, and on the official service marketplaces of Douyin, WeChat Channels, Xiaohongshu and Kuaishou it had long been at the top. Yet growth had been stuck for years — not for lack of effort, but because the channel had never caught up.
The hard part: this industry has zero entertainment value. Nobody kills time scrolling a customer-service company's short videos, so the money he poured into them bought views and conversions that went nowhere. Real clients take a different path entirely: the moment the need appears, they search. And on that one path that actually mattered, what they found was never him.
The turn
We didn't follow the field into the short-video race — for this industry, that road doesn't exist. We put the bet where his clients actually go: search. And we rebuilt "being found" from the ground up, with AI.
- Start from what clients actually type. Pin down the real questions a buyer with a live need puts into the search box, and derive every piece of content backwards from them.
- Four sub-brand sites, holding multiple positions. Not one site fighting alone — several sub-brands each claiming a spot in the results, with link distribution feeding the authority back up.
- A pure-AI content production matrix. From writing to publishing, the whole GEO content pipeline runs on AI — filling every position worth holding, at a volume no human writing team could ever match.
- Watch the rankings, keep iterating. Track the generative-search rankings, patch whatever slips, and hold the way up to first.
Now
Now when a client searches, the answer served first — the one generative search hands over directly — is him. A brand-new acquisition channel has grown out of it: no short-video ad spend, no racing to the lowest bid. Clients come to him, and the deals that close carry six-figure contract values.
More importantly, this is not a one-off campaign. The entire AI content matrix was handed over to the client to run himself — producing, iterating, on its own. We left; the channel keeps growing.
The other half of this company's story — how the management layer itself was armed with AI: The frontline goes to subcontractors. Management goes to AI.