What to Install in Your First Two Weeks

You have seen this company before. Different sector, same shape: the founder is the memory system, decisions queue behind one person, handovers drop things, and everyone is busy in a way that does not add up to progress.

You have ninety days to be obviously worth the money, and the temptation is to arrive with a framework. Resist it. A framework installed before you understand the failure is a solution looking for its problem, and the team will treat it that way.

Week one: measure, install nothing

You are not being paid to observe, so this feels wrong. Do it anyway. Everything you install in week two will be aimed at a number you took in week one, which is also how you prove the value later.

Sit in every recurring meeting. Say nothing. Note two things: how many items are operational versus structural, and how many end with a named owner and a date. Most groups discover their meetings produce discussion and very little ownership.

Count the founder's interruptions. Ask them to tally every question only they can answer, for five days. Twenty is heavy. Sixty means there is no memory outside one head. See When the Founder Is the Memory System.

Ask everyone the same three questions, individually, in private:

• What are you waiting on right now?
• What do you do that you think is pointless?
• What breaks most often?

The third answer is your work list. The second is your credibility: kill one pointless thing in week two and the team will help you with everything else.

Find the commitments nobody has written down. Contracts, renewals, promises made in meetings. Ask for a list; there usually is not one. Assembling it is often the single highest-value artifact you produce all quarter.

Week two: install exactly three things

Not a system. Three mechanisms, each aimed at a number you measured.

1. Split governance from operations. The single highest-leverage change, and it costs nothing. Structural problems are permanently last on a mixed agenda and permanently deferred, which is why the same operational fires recur — the structure producing them never gets a slot. See Governance and Operations Meetings.

2. Write down who decides what, for the ten decisions that actually queue. Not an org chart. Ten lines: this decision, this person, up to this limit. You already know which ten from your interruption count. This is where a fractional operator earns their fee, because you can see the ambiguity the team has stopped noticing.

3. Start decision records, from the next decision forward. Never retroactively — that project dies and takes the habit with it. Five fields, ten minutes, written by the decider. See Decision Records.

That is all. Three things you can hold in your head, aimed at things you measured.

What to leave alone

The tools. Everyone wants to migrate to a new system. Migration is six weeks of pain that produces the same disorder in a nicer interface. Fix ownership and rhythm first; the tool question usually dissolves.

The org chart. Reorganizing before you understand the informal structure moves people around a problem you have not diagnosed.

The founder's involvement in their favorite thing. Every founder has one area they will not let go of, and it is usually the one they are actually best at. Fight that battle in month four, if at all. Spending your credibility on it in week two costs you the other twelve.

Culture. Not because it does not matter. Because it is downstream: it moves when decisions and ownership move, and directly attacking it reads as an outsider criticizing what people care about.

Proving it worked

Re-run week one's measurements at day sixty. Same tallies, same questions.

The interruption count should be visibly down. This is the number founders feel most.

Decisions with a named owner and date should be up. Count them in the meeting.

Items deferred more than three times should be near zero, because the structural ones now have a place to go.

"What are you waiting on?" should get shorter answers.

Bring the before and after to the ninety-day conversation. Most operators describe what they did. Numbers are what turn a contract into a renewal, and a renewal into a referral.

Where it goes wrong

Installing a whole framework in week one. It gets adopted for a month, then abandoned when it collides with reality, and it will be much harder for the next person to try again.

Becoming the new bottleneck. If questions now route through you, you have relocated the problem and made yourself unremovable, which is not the same as valuable.

Fixing what is visible rather than what is expensive. The messy inbox is visible. The undocumented renewal that auto-renews in March is expensive.

Leaving nothing behind. Your work should survive your exit. If the mechanisms only run because you run them, you have been an employee with a strange contract.

What good looks like

At day ninety the founder has fewer interruptions, the team can name who decides what, and there is a written record of every significant decision since your second week.

And when you leave, none of it stops.

Try this week

If you are already inside a client, run the three questions today. Individually, in private.

The second answer — what do you do that you think is pointless? — is the one that buys you permission for everything else. Kill one of those this week.