Douyin campaigns on Qianchuan and Juliang Engine

Douyin creative has a working life measured in days. At meaningful spend we see click-through decay to about half its launch rate within four to seven days, and the accounts that struggle are almost always refreshing creative monthly against a calendar rather than retiring it against its own curve. Our operating rule is a three-day rolling threshold: when a video falls below sixty percent of its first forty-eight hours, it comes out, whatever it cost to produce. Which engine, and why the choice is not cosmetic: Qianchuan is for advertisers with a Douyin store and a reason for the viewer to buy inside the session. Juliang Engine is for app installs, lead capture and anything transacting elsewhere. They price differently, optimise against different events and reward different creative. We pick one as the primary engine at the start and state what adding the other would cost later. Running both from month one is how a budget becomes two underfed halves that never leave the learning phase. The store and the live cadence come first: We do not take Qianchuan work from an advertiser with no store and no intention of streaming. The strongest inventory on that engine is the live room feed, and a room opening twice a month cannot hold a bid against rooms opening five days a week. If streaming is not going to happen, the honest recommendation is Juliang Engine with a landing page and a lower ambition on cost per order, and we make that recommendation before you sign rather than in month two. Creative supply: We commission and direct short-form video with mainland creators through Xingtu where a face is needed, and hold the script, hook structure and first-second framing ourselves. Volume is set by the decay curve, not by a number agreed in advance. Each cycle tests one variable, hook or offer framing or presenter, because three variables at once produce a winner nobody can explain and therefore nobody can repeat. Account structure and bidding: Flat structures, few ad groups, high spend per group. Douyin's optimiser wants event volume more than it wants your segmentation theory, and the twelve-group account built to mirror an org chart is the most common thing we dismantle in week one. Manual bid floors are used only where a cost ceiling is contractual, and we will tell you what that ceiling costs in lost volume before you impose it. What we refuse: Buys measured on video views. That number is one the platform is happy to sell and it has never predicted revenue in an account we have run. Follower campaigns with no downstream reason for the follow. Any creative that puts a child in the seller's position, and any live cadence aimed at an under-18 audience; Douyin's minors rules are stricter than most European advertisers expect and enforcement lands at account level, not at campaign level. Not included: Store setup, product listings, logistics and after-sales. Hosting or scripting your own team's live sessions, though we will watch recordings and give notes on pacing and offer placement. Anything that would put us in charge of the inventory whose performance we are also reporting on. Who this is not for: Advertisers who want a brand campaign with a reach target. Douyin will happily sell you one and we are not the shop to buy it on your behalf. Reporting: Weekly through a build cycle, fortnightly after. Spend, paid ROI as the engine reports it, order data from your own store export, the decay curve for every live creative, and the retirement decisions taken that week with the reason attached to each.