Launch week for an integration

Shipped code is not a launch. Listing, help article, seller sentence, the customers who asked, and someone watching errors.

Written lesson. Do the exercises in order. Put the artifact in your hiring packet.

A connector that goes live on a Friday with no listing and no email is a tree falling in a forest. Launch is a project with a week, not a git tag.

The checklist

  • Listing live, or a dated reason it is not.
  • Help article: connect, first job, what to do when it fails.
  • Seller sentence and the one-pager, in the place sellers already look.
  • Personal note to every customer and prospect who asked for this by name.
  • CS enabled: they know the setup time and the support path.
  • Partner notified, with a link they can send to their sellers.
  • Monitoring: error rate, auth failures, a human on-call for week one.
  • A 30-day check on active connections, already on the calendar.

Each line has an owner. If the owner is “someone,” it will not happen.

Who you tell

The people who already wanted it, first. Then your list. Then a modest public note. A giant campaign for an integration nobody asked for is how you learn, expensively, that pull was fake.

The day after

Watch setup drop-off. If people bounce at OAuth, you do not have a marketing problem. You have a setup problem. Fix that before you write another tweet.

QBR already booked

Launch without a follow-up date is how launch-and-forget starts. Put the first review on the calendar before you celebrate.

If you can only do one thing besides the code, do the customer notes. Those people are the first reviews and the first installs.