The 90-Day Install

Ninety days is the first test of the operating method, not a guarantee that every business result will arrive in one quarter. It extends the first-month plan in the book's Conclusion: days 1 through 30 run the same four weeks, using a meeting already on the calendar as a weekly review between the formal reviews, and days 31 through 90 widen the trial and verify the result. Module 1 finds the dependency. Module 5 puts the initiative on the record, the course's version of the shared record in Chapter 5. The windows below follow the work, and every other module is pulled in only when its problem appears. A facilitator runs the install; the outcome owner has work to do in every session and should not also be running the room.

This lesson walks the fictional onboarding example from kickoff to day 90. Not a result from one of my companies; the names are role labels. A business onboards 20 new customers a month, each using an average of 12 staff hours and taking 10 business days. The team wants 90 percent of the next 30 consecutive onboardings, at least 27 of 30, completed within seven business days by day 90, while meeting its existing accuracy and customer acceptance standards. Before the work starts, the team fixes the first case in that set and the events that start and stop the clock; difficult cases are not excluded.

The Hero Test found the dependency: every order with nonstandard terms waited for the head of customer operations to approve the configuration personally. That person owns the outcome. Sales operations owns complete order data, finance owns credit approval, and delivery owns configuration, with a written rule for the standard exceptions and a named backup for the judgment calls, coached by the head of customer operations through a real case and an exception. The sponsor delegates up to $10,000 of spending and the authority to change the internal sequence, and reserves customer contracts, legal controls, and spending above that limit, each with a named approver and a two-business-day response commitment. The first commitment is small on purpose: by day 5, sales operations and delivery agree on the required order fields and check one recent case against them.

On day 90, the account manager who takes over each customer after onboarding stands in for the customer as the receiving owner and verifies the result against the acceptance standard, and the sponsor confirms the evidence. Faster onboarding does not count as success if errors rise or customers cannot use the service. If average staff time later falls from 12 hours to eight, 20 onboardings release 80 staff hours a month. That is capacity, not automatically cash. By day 90 the approval that waited on one person has become a rule the team applies and a backup who can make the judgment calls.

The facilitator collects three things: the 48-Hour Decision Queue Audit from Module 1, run over the two working days before the session; the current commitments and the people or money they consume, from Module 6; and the baseline for the outcome, in the terms of the Outcomes line. Before the session, the facilitator asks the people doing the work what sends it back to the same person, and whether the cause is missing authority, resources, knowledge, access, or practice. The facilitator also confirms that a sponsor who can grant authority will be in the room, or names who will carry open questions to that person and by when.

The first session runs Sessions 1 and 5 together. Open with the Hero Test, each person answering for themselves before anyone names a capability the company waits on. Run the Bottleneck Check and sort the audit's entries into the three groups: decisions that belong at the top, decisions an existing owner should make, and decisions nobody has been authorized to make. Choose one recurring dependency to change. Draft the front page of the VOOCS Operating Record, and write the Authority Grant for the new owner, including the exceptions that need another decision, and the Done Test for the result in that record. Commit the resources, name the backup, and have the people who control those commitments confirm them. Put the reviews on the calendar. Leave with a decision or a specific evidence request, an owner, a date, and a way to check the result.

If no one present can authorize the decision, record who can, who will take it to them, and the date it must close. That becomes the first open entry in the Waiting Decision Log, and the install starts anyway.

If the sponsor cannot grant the authority the work needs, or the capacity disappears, record that as the finding and escalate or stop rather than asking the owner to absorb it. When a commitment is missed, decide what condition must change before setting another date.

Do not declare success because day 90 arrived, or because a week passed quietly. A smaller queue can hide stalled work. Released hours can sit unused. Documented procedures can leave judgment in one person's head. Record those results honestly and choose the next action. The system is improving when the work improves and other people can carry it.