The worst first 90 days look productive. You meet everyone. You sign three logos. You put a partners page live. Nothing a customer can use has shipped, and product has already stopped taking your meetings.
Do this instead.
Days 1–30: the map
List every company already called a “partner.” For each one, write four fields: shared customers (named), workflow (one sentence), commercial motion (or none), shipped artifact (listing, connector, or none). Most of the list will fail the last two.
Talk to sales, CS, and five customers. You are mining “we export this to X every Monday,” not collecting opinions about which logo would look nice.
Output of month one: a scored list, a recommended kill list, and one partnership you will take through to shipped.
Days 31–60: one shippable relationship
Pick the high-pull, high-distribution name. Get a named owner on their side. Send a one-pager. Run a discovery call. Write the scope: user stories, data ownership, acceptance criteria.
In parallel, clean the rest. Email the dead ones a clean “we are pausing.” Do not ghost them. Partner managers move companies. The email is cheap insurance.
Output of month two: a scope an engineer can build, a date, and a quieter book.
Days 61–90: launch as a project
If the build is not done, your job is to keep it from slipping, not to start partnership number two. When it ships: listing, help article, sales one-pager, personal note to every customer who asked. Then a 30-minute QBR scheduled before you celebrate.
Output of month three: one live, adopted integration and a cadence. That is a first quarter a CEO can understand.
What you report
Not “meetings held.” Report: partners killed, partners in scope, one ship date, first installs. If the CEO wants more logos, show them the matrix. Volume without a workflow is a tax.