Switching factories is expensive in time, not necessarily in money: 8-14 weeks for a custom product, 3-5 for stock goods. The gates that justify a switch are capability failures (repeated failed PSIs after a reviewed CAP, concealed subcontracting, ownership change that erased your contacts), not first-round price gaps. The mechanics that save the season: qualify the replacement on production tooling before anything announced, run dual-track with the new factory at 20-40% volume for 2-4 weeks, migrate the eight checklist items (tooling with release report, sealed samples, certs, packaging plates…), and leave the old factory with a clean invoice trail — the industry is smaller than it looks.
Switch or Fix? The Four Decision Gates
Buyers swing between two unhealthy poles: staying loyal to a failing factory for years, or threatening to leave every quarter until the threat devalues. The four gates keep the decision honest:
| Gate | Switch trigger (capability) | Fix trigger (relationship) |
|---|---|---|
| Quality | Two consecutive failed PSIs after a jointly reviewed CAP; concealed spec or material breach | Single-batch drift with a clean root cause ("new resin lot", "new line operator batch"); CAP executed on time since |
| Price | Gap ≥ 15% verified by triangulation AND refused twice in writing with volume commitment | Gap < 10%; or offset-able via the MOQ and spec levers (packaging, carton spec, decoration tier) |
| Capacity & reliability | Two missed sailings in a quarter; order-book at 110% and declining responses; confirmed utilization red flags | One missed sailing with fair warning and a credible recovery plan |
| Entity & continuity | Ownership sold/bought with your contacts erased; unverifiable subcontracting discovery; license revoke risk | Sales rep rotation (re-map the authority chain instead); minor name-change with same beneficial owner |
One asymmetry worth internalizing: the credible option to switch is worth more than most switches. Buyers who maintain a qualified, audited second source get better prices from their primary without ever moving — the negotiation playbook covers that leverage math in the triangulation section. A switch you execute is expensive; a switch you could execute tomorrow is negotiating power that pays rent.
Qualifying the Replacement: The 21-Day Sprint
The worst way to qualify a replacement is announcing it ("we're considering a second source") — you'll get showroom prices, priority scheduling theater, and a primary supplier who suddenly finds quality problems in your files. The sprint below runs under cover of normal new-project traffic:
- Days 1-3 — list and screen. Three candidates from the relevant category cluster; run the free 10-check verification routine on each. Kill any that fail checks 2-4 (license scope, tiny-custom-order test, address reality).
- Days 4-7 — video audits. Unscripted line walks for the survivors; the ladder tier $0. If your tooling might transfer, ask one question here that changes everything: "can your press shop run a 2-plate cold-runner mold from another maker?" — willingness signals toolroom maturity.
- Days 8-14 — on-site audit. $150 on-site for the finalist (clustered with other factories in the zone if you have them); register anything tooling-shaped per the audit add-on.
- Days 15-21 — PPS on production tooling + capacity math. Paid PPS per the golden-sample protocol, plus the honest capacity question: their order-book fullness vs your delivery dates. A replacement at 110% utilization is not a replacement.
Gate at day 21: sign nothing yet, but you now hold a qualified, audited, sampled Plan B with a written quote. That document changes your position with the primary factory even while you run the fix path first.
The Dual-Track Transition: Split Orders in Parallel
Cold-turkey cutover — last order at factory A, first order at factory B across a quality-validation gap — is how buyers lose a season. The dual-track transition splits risk correctly:
| Phase | Weeks | Factory A (old) | Factory B (new) |
|---|---|---|---|
| Parallel start | 1-2 | Current order proceeds at 100% of its normal share | First PO at 20-30% of volume, non-critical SKUs or the simplest product variant |
| Ramp | 3-5 | 60-70% share; starts seeing your B-side quality data if asked (share selectively) | 30-40% share, first PSI run, first seal-vs-mass-production comparison |
| Flip | 6-8 | 30-40% share, or holds buffer stock only | 60-70% share, second PSI clean |
| Wind-down | 9-14 | Final settlements, tooling release if migrating, clean invoice trail | 100%, stabilization runs logged |
Two rules keep dual-track honest: never tell either factory the full transition plan (the primary will deprioritize you the moment your exit is certain — keep them hungry until the wind-down phase; the new one will hide problems while your volume is small). And both PSIs run at full standard — a "small order exemption" on the new factory's first batches is how transfer-attributable defects reach your customers.
The Migration Checklist: What Actually Moves
The eight items that physically or legally cross the factory boundary. Print this; the tooling playbook's clause checklist covers the ownership paper, this list covers the physical and legal migration:
| # | Migration item | Status |
|---|---|---|
| 1 | Tooling: release report + per-plate photos + serial verification against your ledger; insured skid freight | |
| 2 | Golden samples + seal records: your half travels to the new factory's line for the T1 comparison | |
| 3 | Spec sheets + revision history + the PPS photo archive (material declarations, color chips, test reports) | |
| 4 | Certificates: re-verified under the new factory's name if cert holder changes (CE/UKCA/FCC procedures restarted at day 1 — the cert timelines apply) | |
| 5 | In-transit and warehoused stock you own: counts verified, ownership papers matching the PIs | |
| 6 | Packaging plates, print files, dielines, artwork with font licenses | |
| 7 | PI templates with your clause blocks (ownership, sealing, PSI, payment ladder) — pre-loaded, not re-negotiated | |
| 8 | Contacts ledger replacement: new factory's authority chain mapped (the field guide's who-decides routine) before first crisis |
Item 4 is the one that slips every schedule: certificates follow the legal manufacturer, and a mold that transfers in 10 days can drag 6-10 weeks of recertification behind it — for regulated categories, start the cert transfer conversation before the PPS sprint finishes.
The New-Factory Ramp: Stabilization Without Surprises
The first 2-4 runs at factory B will differ from factory A's output even with the same mold, same golden sample and same spec sheet: thermocouple calibration, cooling-water temperature, line rhythm and operator technique all live in the factory, not the tool. The ramp discipline:
- T1 first-article against the seal, not against memory. The new factory's first-article check compares to your sealed golden sample and the photo ledger — never to "how factory A used to make it". The seal wins; that's what it's for.
- Expect a defect-rate hump, measure it. Factory A's steady-state PSI defect rate of ~1.5% will read 3-6% in runs 1-2 at factory B — most transfer-attributable (cosmetic handling, packing, color temperature interpretation). Classify every defect in runs 1-2 as transfer-noise vs true-fail before reacting; only true-fails count against the switch decision.
- Front-load DUPRO. During the during-production stage, both factories' lines deserve in-line eyes while they coexist — duplicating DUPRO at $150-350 per run for one transition cycle is cheaper than one container of drifted goods.
- Log everything for the second transition you won't have to run. The completed checklist, T0/T1 reports and ramp curves become your playbook — the second switch (there's always a second) takes half the time because the muscle memory exists.
The Exit Itself: How to Leave Cleanly
How you leave determines how the next three factories treat you — and whether your tooling, prints and pending payments leave with you:
- Settle the invoice trail to zero. Pay the final balance on the PSIs that passed; document (photos, timestamps) every dispute you're waiving or discounting by paying. A clean settlement is cheap reputation insurance in an industry where the good factories all know each other — the field guide's reputation note applies to exits.
- Release the paperwork ghosts. NDA/NNN clauses survive the relationship; so do your mold-registration records. Request written confirmation that no tooling, artwork or tooling-derived files retain your IP and that your designs won't be offered in trade-show showrooms. Factories sign it when asked on a settled account.
- Keep the bridge, not the obligation. The closing line that preserves optionality: "the door stays open next season." Yiwu-scale manufacturing collective memory is long — a buyer who leaves politely at scale 20k and returns at scale 200k gets the red-carpet quote.
- Never exit during a held container. If goods are in production or shipping, the exit conversation waits until landing and PSI — leverage you hold mid-shipment reverses the moment you need their paperwork cooperation for customs and the documents chain.
Frequently Asked Questions
When should I switch suppliers instead of fixing the relationship?
Switch on capability: two consecutive failed PSIs after a jointly reviewed CAP, concealed subcontracting, an ownership change that erased your contacts, or a safety-critical breach hidden from you. Fix with one structured warning: price gaps under 10%, single-batch drift with a clean root cause, or communication problems that respond to a channel change. The credible option to switch is worth more than most switches — qualified second sources improve your primary's pricing without you ever moving.
How long does a supplier transition take?
Custom products: 8-14 weeks end-to-end — 2-3 weeks to qualify (video audit → on-site audit → PPS), 2-4 weeks dual-track overlap at 20-40% new-factory volume, 2-3 stabilization cycles, then wind-down. Stock or semi-custom goods compress to 3-5 weeks because the PPS stage collapses into shelf verification. The binding constraint is usually recertification for regulated categories, not production.
What do I need to move to a new factory?
Eight items: tooling (with release report and verified serials), golden samples and seal records, spec sheets plus revision history, certificates (re-verified under the new legal manufacturer), your in-transit and warehoused stock, packaging plates and print files, pre-loaded PI templates with your clause blocks, and the new factory's authority-chain map. Certificates are the long pole — start their transfer before PPS finishes.
Will my current factory find out I'm qualifying a replacement?
Only if you tell them, use their referrals, or show up at their competitor's gate with the same FOB sheet in your bag. Screen candidates under normal new-project framing ("new SKU line we're evaluating"), pay for your own audits, and bring clean-spec questions rather than comparative quotes to any factory visit. Factories assume dual-sourcing exists; assume they'll treat your data as shared the moment it's interesting.
Is dual-tracking unfair to either factory?
It's standard practice above $50k annual volume and both factories assume it exists even when it doesn't. What's actually unfair: running one factory's engineering time against specs you've already decided to move, or promising volume you've already scheduled elsewhere. Keep both PSIs at full standard, share phase-appropriate information, and never let the primary discover your second source from a courier or a mutual booth neighbor at the fair.
My factory's prices jumped 12% on re-order — switch now?
Run the gates first: verify the increase is real across the triangulation quotes (raw material indices move everything — check the real-prices data before assuming padding), then counter with a volume commitment or the spec levers. A 12% jump with no material-index cover is a padding play deserving one written warning; two refusals later, your day-21 qualified replacement converts the standoff into a 4-6% settlement — or a smooth executed switch.
Weighing a switch right now?
Send us the gate evidence (PSI reports, price trails, what changed at the factory). We'll score switch-vs-fix against the four gates and flag the migration list for your product — free, within 48h.
Request Switch Assessment →📍 Make this the next step, not the last