Customer Onboarding Plan Template

Lincoln Murphy, a customer success consultant, warns vendors not to confuse their definition of success with the customer's. His advice: if you don't know what success means to them, ask¹. He suggests asking how they measure success and how their boss measures them.

Good answers describe a moment in someone's week, not a broad goal. Turn it into an event two people would agree happened, with a witness on their side, since usage data can't show whether anyone helped. Record your activation event, the product milestone you track, separately: your consultant can trigger it before the customer gets any value.

Signature and kickoff are your dates. Ask for theirs: a peak season, an audit, a store opening, the day their old contract ends. Some are fixed by law. A US federal agency's fiscal year ends on September 30², however late you kick off.

Then test it: ask what happens if it slips and whether they've told their own customers. A date their business runs on rarely moves. Count every milestone in days before that deadline, not days since signature.

If there's no date, write “none” rather than borrowing your quarter end. Without one, nothing in their business pushes their side along.

Much of what decides the timeline sits on the customer's side, and you can learn it in week one. Ask out loud at kickoff: a form sent afterward gets “we'll assign an owner” instead of a name and a share of someone's week.

Mark each factor clear, friction or blocker. A blocker stops other work, like a security review that must finish before you can set up their account. Friction only costs days, like data in three systems.

If their review hasn't started, send your security documents, such as a SOC 2 report, within days. The AICPA, which sets the standards for these reports, says they give customers the information needed to assess the risks of outsourcing³. A security questionnaire template helps with any questionnaire they send.

Don't pick 30, 60 or 90 days because they're round. Bent Flyvbjerg, a project management researcher building on Daniel Kahneman's work, explains one big reason plans run late. People focus on the plan in front of them⁴ rather than on how similar work turned out. The cure he describes, reference class forecasting, predicts from what comparable past projects actually took.

1. 1

   Start from a base

   Use your fastest recent onboarding of a similar-sized customer where every factor was clear.

2. 2

   Add what each risk cost

   For each factor that isn't clear, add the median extra days it cost past customers in the same state.

3. 3

   State the sample

   Note how many past onboardings each number rests on. Below three, call it a rough estimate.

4. 4

   Explain any outlier you drop

   Exclude an unusual project only when its known cause won't recur here, and write the cause down.

Work back from their event and their date, the way Murphy suggests backing out from initial success¹ to the milestones before it. Give each milestone a side (yours, theirs or both), a named person rather than a team, and a date.

Mark the critical path, the chain of tasks where any delay moves first value. It often runs through their work, so count their tasks out loud at kickoff: a plan showing only yours is a promise you can't keep alone.

Push anything they want but don't need for first value, such as extra integrations, past their date, and agree that in writing. If sales used a mutual action plan, carry its named contacts into this one.

Agree escalation while nobody is annoyed. UK government project guidance calls the agreed leeway tolerance⁵: how far work may drift from the plan before it goes up a level of management. Set yours now, with one contact per side at each level and triggers such as a critical-path task more than three working days late.

First value from one person is only a start. List everyone meant to use the product, by name, and track each one:

• Independent: has done the core task alone, more than once
• Assisted: has done it, but only with help
• Trained, not started: trained, but hasn't done it yet
• Not started: has never logged in
• Viewer: looks at reports, often the buyer; not adoption

Before offering more training, find out why someone never logged in: some were never invited, and some joined after the plan was written. If only one person can do the core task alone, train a second before widening the rollout, or one resignation takes the account back to zero.

When the witness confirms their event, add a row to your onboarding history: readiness at kickoff, predicted and actual days, and one sentence on what cost the time. Then carry the value definition forward as a goal to score at each quarterly business review.

This template has a file for each step. With Edit with AI, River takes the readiness read from your kickoff notes and sizes the plan from your past onboardings.