The playbook
A 30-minute kickoff. A working configuration. A real workflow in production. That is the whole playbook, and it fits on an index card on purpose.
The rule behind it: nothing is "set up" until a real person has done a real piece of their real job in it. A configured workspace with no traffic is not onboarding, it is furniture.
Hour 0 — kickoff call, one decision: which workflow goes first.
Hour 1 to 4 — we build that workflow with them on the call, not after it.
Day 2 — they run it themselves, we watch and say nothing unless asked.
Day 5 — a 20-minute review, and the second workflow gets scheduled or dropped.
Hour zero to hour four
The kickoff has one job: pick the smallest workflow that someone is doing today, by hand, at least weekly. Not the most valuable one. The most repetitive one.
Then we build it live. This feels inefficient and it is the highest-leverage part of the whole process — the customer sees which decisions were arbitrary, which is exactly the knowledge they need on day 30 when they want to change something.
We do not import historical data on day one. It is the most requested thing and the most reliable way to spend three weeks before anyone has used the product.
Day two, and the part that usually goes wrong
On day two we watch and stay quiet. Every instinct says to jump in when they take the long route through a screen. Staying quiet is how we find out which long routes exist.
The most common failure is not technical. It is that the person on the kickoff call is not the person who does the work. When that happens we stop and reschedule — the two-day timeline is not worth the six months of a tool nobody adopted.
The day-five review
Twenty minutes, three questions: what did you stop doing, what did you start doing twice, and what are you still doing in the old system.
The third question is the one that matters. A process still running in the old tool on day five is a process that will still be running there in a year, and it is almost always a sign that we picked the wrong first workflow.
If the answers are good, the second workflow gets scheduled. If they are not, we redo the first one rather than adding a second. That rule costs us a week roughly one time in five and has never been the wrong call.
When it fails
Three failure modes, in order of frequency. The kickoff has the wrong person in the room. The chosen workflow needs an integration that is not ready, so day two becomes day nine. And occasionally the team is being migrated by someone above them who has not told them why.
The third one is not a process problem and we have stopped pretending it is. When we smell it, we ask for fifteen minutes with the people who will do the work, before anything gets configured.
Why rigour beats tooling
The tool matters less than the rigour. We replaced a 6-step internal CRM sequence with a shared document and got better outcomes, because the document was read and the sequence was clicked through.
Onboarding is not a configuration problem. It is the process of finding out which of your assumptions about the customer's job are wrong, as fast as possible.