Lead times from China: what's real and what's optimism.

The five clocks

First, pre-production: final sample approval, materials ordered and received, line scheduled, a clock that hasn't started until the deposit lands and the golden sample is signed. Second, for new products, tooling: weeks of machining and trial shots before unit one exists. Third, production itself, usually the most honest number in the quote. Fourth, quality and pack-out: inspection, any corrective loop, cartonisation. Fifth, freight and clearance: the mode you chose plus port handling and customs on both ends. Digital trade documents can take some of the waiting out of clearance, but only when the paperwork is complete and correct. "30 days" quotes typically describe clock three. Your launch date lives at the end of clock five.

Where schedules actually slip

Rarely on the production line. The classic slips: material lead times nobody surfaced ("the resin arrives in three weeks"), your own approval latency (sample sign-offs and drawing questions that sit in someone's inbox count against the clock too), holiday compression around Chinese New Year and Golden Week, when factories drain and refill over weeks, a failed inspection triggering an honest corrective loop, and peak-season freight, where vessels roll bookings exactly when everyone's stock is due. None of these are surprises to people who've shipped; all of them are surprises in first-timer schedules.

The buffer strategy that keeps launches honest

Plan on the full five-clock sequence, add buffer proportionate to novelty (new product, new supplier and new tooling each deserve their own margin), and set the public launch date off the buffered clock, not the brochure one. Then manage the schedule actively: milestone checkpoints tied to payment gates, photos and evidence rather than reassurances, and the discipline to hear about a slip in week two, while it's recoverable, instead of week ten. If the date is immovable, the buffer converts to a plan: a small air tranche protecting launch while sea carries the volume.

The agent-native lens: watching all five clocks

When the buyer sends an assistant. The lead time an AI assistant finds for a product made in China is most likely clock three, because that's the number quotes and factory pages publish. Ask for each clock separately instead: what starts it, how long the supplier expects it to run, and what it depends on. Have the assistant lay the plan over the holiday calendar and your own approval steps, and mark every date that rests on a reassurance rather than a confirmation, such as a material order, a tooling trial or a vessel booking. The same discipline applies when you're the one quoting. If you sell to other businesses, a lead time given as a dated estimate with its assumptions is one a customer's assistant can plan around; a bare "30 days" invites exactly the optimism this article warns against.

When the business runs on agents. The clock agents help with most is often your own. Of the slips above, approval latency is the only one inside your walls, and an agent can chase internal approvers as readily as suppliers and report how long each decision waited. Across orders, agents can record quoted against actual durations for every clock and supplier, so the next plan's buffer comes from your own history rather than a rule of thumb. People still set the public launch date, decide how much margin novelty earns, and decide when the buffer turns into an air tranche. The agent's job is to make sure that call happens in week two.