The friction list

Your map follows one piece of work from one point of view. The friction list collects everyone's. It's where the people who do the work write down where it waits, piles up or gets redone, week after week, so the bottleneck can be found from evidence rather than from whoever argues best. It's also the most fragile thing in the framework, because it only works if people trust it.

What goes on it

Three kinds of friction, the same three you looked for on the map:

• Work that waits. “Quotes wait about six days for pricing review.”
• Work that piles up. “Reconciliations pile up in the last week of the month.”
• Work that gets redone. “One in four proposals is sent back for missing figures.”

Anyone can add to it, from the CEO to the newest assistant. Items arrive from other places in the framework too. When something is raised at the front door and parked, it goes on the list. When something goes wrong with AI, the “learn” step of stop, tell, fix, learn leaves a short blameless note here. Later, when anything that runs reaches its review date, the findings land here as well. Over time the list becomes the organisation's memory of where work gets stuck.

Rework deserves a special look in an AI rollout. AI makes it cheap to produce work that looks finished and isn't, and the cost lands on whoever receives it.

Record the work, never the person

This is the rule that makes the list work. Every item describes what happens to the work: where it waits, how long, how often it comes back. It never describes what someone is like.

There are two reasons. The first is practical. “The finance lead is too slow” can't be fixed; nobody knows what to change. “Invoices wait five days for finance sign-off” can. It has a place, a size, and a test: did the wait get shorter?

The second is about trust. A list that names people reads as surveillance. The first time someone sees their name on it, they stop adding to it, and so do their colleagues. When the Deployment Framework was red-teamed, this was one of the reasons it fell short on the fear test: would a worried employee still tell the truth? The rule to record the work, not the person, was the fix.

Deming was estimating where problems and chances to improve come from, and his answer is why this rule isn't only kindness. Most friction comes from how the work is set up: handoffs, rules, batches, missing information. Blaming the person standing nearest to it hides the cause.

Every item has a visible status

Every item carries one of four statuses, and everyone can see them:

• New: raised, not yet looked at.
• In progress: someone is working on it.
• Parked: seen, and not this loop. It stays on the list, with a reason.
• Done: fixed, and checked.

The status is what keeps people contributing. A list where items vanish, or sit at “new” forever, teaches people that adding to it is pointless. A list they can see moving, including items honestly parked, teaches them it's worth the effort. Parked isn't a polite no. It's an answer.

Start yours below. The piles from your map in lesson 1 can come straight across, and a few examples are there if you want to see the shape first. If something you type reads as if it's about a person, the builder will say so and suggest a rewrite.

Go through it every week

The friction list belongs to the weekly meeting. Every team, every week, fifteen to thirty minutes, inside a meeting it already holds. The agenda is short: go through the list and its statuses, raise any problems with AI output, and check which reviews are coming due. Managers bring the team's friction to the surface; everyone adds their own.

In days 1 to 14, aim for two weekly reviews before you name the bottleneck. The first fills the list. The second shows which items keep coming back. Where friction clusters, at one step or one handoff, is where to look hardest in the next lesson.