How to switch white label partners without losing the client

One late launch is a conversation. Repeated missed QA, disappearing seniors, surprise invoices, and brand leaks are a pattern. Write the pattern down with dates. You will need it for the incoming partner and, if it gets legal, for the MSA. Feelings are not an exit clause. Common white label mistakes are often why you are here.

You cannot migrate what you cannot list. Before notice, quietly inventory: repos, package registries, hosting, DNS, third-party apps, CI, design files, ticketing, staging URLs, production credentials, analytics, and who is the billing owner. If that list lives only in the partner’s head, you are already late.

If you signed a decent MSA and IP clause, transfer is a right. If you did not, you still need the list. Negotiation without a list is begging.

Pay what you owe for accepted work. Do not steal a sprint. Do not smear them in a client Slack. You are allowed to be firm about access. You are not allowed to be chaotic. Incoming partners watch how you treat the outgoing one. Clients can feel the chaos even when they never hear the name.

The agency that ghosted their last partner ghosted us on invoices six months later. We should have treated that as data.

Hard cutover (Friday they stop, Monday someone else starts) works for a brochure site with no pipeline. It fails for retainers, commerce, and anything with weekly releases. Parallel run means the incoming partner has read-only then write access while the outgoing partner finishes named tickets. You pay overlap on purpose. Overlap is cheaper than an outage.

Name a freeze: outgoing partner does not start new features after date X. They only support production and transfer. Incoming partner does not “improve the architecture” during overlap unless you sold a rebuild. Revision discipline applies; a switch is not a free redesign.

Give them the inventory, the stack, the clients in flight, and the last two months of tickets. Do not give them a novel about how terrible the last people were. They will spend energy gossiping instead of reading the repo. Use the brief template plus a cutover exhibit.

1. 01

   Access and clones

   They can build staging from what you actually have, not from a deck.

2. 02

   Shadow a release

   They watch one production change with the outgoing team still on the hook.

3. 03

   One contained ticket they own

   Prove QA and communication before you hand them the flagship.

4. 04

   Admin transfer complete

   You, not they, hold org owner. Then the outgoing partner can leave.

Score them with the same scorecard you should have used the first time. Switching is not an excuse to skip vetting because you are in a hurry. Hurry is how you repeat the year.

Do not start with the angry enterprise account and the store that takes weekend orders. Start with a low-blast-radius site: brochure, staging-heavy, or a retainer with a calm AM. Prove access transfer and a release. Then move commerce. Then move the client who already asked why Friday deploys feel haunted.

If two clients share a monorepo, they are one move. Pretending they are independent is how you break the quiet one while saving the loud one. Map blast radius on a whiteboard before you send notice. Overflow and capacity thinking helps: you only have so many careful hours in a week. Spend them on sequencing, not on a manifesto.

If they never knew a partner existed, do not announce a “vendor change.” Announce nothing, or announce a process improvement if something will be visible (status cadence, staging URLs). If they knew, tell the truth without a villain: you are changing production capacity to protect quality and response times. What clients should know is the policy layer. The switch is a special episode of that policy.

Language that does not create a procurement event

Avoid “we outsourced you.” Prefer “delivery capacity is staying inside our process; some of the engineering roster is changing.” If a named engineer was on client calls, you need a named replacement and a handshake. People notice people. Process documents do not replace a face they trusted.

If they refuse to transfer, escalate with the contract, then with counsel if the sites are real money. Do not try to “rebuild from staging” as a clever trick if you do not have rights—that is how you buy a second legal problem. Pay undisputed invoices. Dispute the rest in writing. Cutting payment while they still hold DNS is how weekends get exciting.

Going forward, never let admin live only with a partner. That rule is in the MSA article for a reason. The switch is the expensive way to learn it.

Licenses in the partner’s company name—analytics, error tracking, premium plugins—are easy to forget and ugly to unwind. Put a license column in the inventory. Transfer or replace before you revoke their cards. A monitoring tool that dies on cutover night looks like you broke production when you only broke a credit card.

The incoming partner needs a staging that is not production with a different hostname guessed in a panic. Clone, secrets, and a smoke test before DNS moves. If staging only exists on the outgoing partner’s laptop, you do not have staging. You have folklore. Build a real one under an account you own, even if it delays notice by a week.

Production cutover should be a checklist with rollback: DNS TTL lowered in advance, health check, form test, payment test if commerce, and a named human awake. QA before client review is the calm version of this list. The switch is the same list with higher blood pressure. Write it while you are calm.

Preview deploys and CI that still point at the outgoing org will betray you a month later when a “small fix” publishes from a pipeline you thought was dead. Hunt webhooks. Hunt GitHub apps. Hunt Netlify/Vercel teams. The inventory is not done until the robots are yours.

1. 01

   Days 1–5

   Inventory, access audit, incoming partner under NDA, notice drafted not sent.

2. 02

   Days 6–10

   Notice sent. Overlap calendar. Freeze on new features. Client language approved internally.

3. 03

   Days 11–20

   Incoming partner on staging. Shadow release. Admin transfer in progress.

4. 04

   Days 21–30

   They own new tickets. Outgoing on-call for production only. Access revoked when checklist is green.

If the inventory does not exist, add two weeks before day 1. Pretending you can skip discovery is how DNS stays in a personal account named after a contractor who left in March.

After cutover, watch the same things you should have watched before: QA leaks, maintenance coverage, and retainer ROI. If the new partner only looks good because the last one was worse, raise the bar before you add clients.

If you want a partner used to taking over live agency work under your brand, look at how Tretanz runs delivery and book a call. Bring the inventory. The useful first meeting is access and freeze dates, not a tour of grievances.