The Interface Ledger

Monday looked green.
Every team met its plan.
But the customer still waited.

The delay wasn’t inside the work, it was between it.
Requests slid across tools, bounced between calendars, paused for “one last review.”
No single owner. No visible clock.
The seam had no ledger.

Quinn’s move: give every interface a receipt. Not to shame, but to surface.
When the work moves, the numbers move with it.

A living register with five sections. Updated weekly. Visible to both sides and leadership.

1) Flow inventory (In / Out)

Items received this week (by lane/type)
Items accepted, rejected, returned
Net WIP at the seam (start-of-week → end-of-week)

2) Queue health

Oldest item age in queue
80th percentile wait time to first commit
Percent of items breaching SLA-to-first-response

3) Handoff latency

Median time: ask → estimate → commit → verify
Percent verified on first pass
Bounce count (times returned before acceptance)

4) Breach receipts

Count of SLA breaches (by clause)
Hours credited/debited (service credits or recovery hours)
Root cause snapshot (capacity, criteria, change, quality)

5) Rework cost

Items requiring rework post-acceptance
Hours of rework attributed to the seam (not the team)
Dollarized cost (standard loaded rate × hours)

The ledger is not a dashboard. It is an evidence log with math.

How the data is captured (low lift)

Entry stamps: When an ask hits the interface path with minimum facts.
State changes: Estimate given, commit made, delivered, verified.
Outcomes: Accepted, returned, or rejected with reason.
Breaches: Auto-flag when timers exceed SLA; create a receipt line.
Rework: Any post-verification fix tied back to the seam ID.

Tooling can be Jira/Linear + a shared sheet. The behavior matters more than the brand.

Weekly cadence (15 minutes)

Before the review: Ops PM closes the ledger Friday 5 pm.
In the room (both owners):
1. Oldest item age > target? Decide the unblock.
2. First-pass acceptance < target? Tighten entry criteria or training.
3. Breach receipts trend up? Apply credits or reduce intake.
4. Rework cost rising? Add/adjust interface tests.
After: Post decisions on the same page. No extra slides.

Targets (tune after two cycles)

Oldest item age: ≤ 5 working days.
P80 wait to first commit: ≤ 3 working days.
First-pass acceptance: ≥ 90%.
SLA breach rate: ≤ 5% of items.
Rework cost: ≤ 3% of seam hours.

What leaders actually manage

Capacity at the seam: If queue age grows, re-balance capacity or narrow intake.
Criteria integrity: Low first-pass means the “door” is fuzzy—fix the criteria, not the people.
Price of misses: Breach receipts convert hand-waving into credits or recovery hours.
Change discipline: If changes drive breaches, shrink windows or raise notice periods.

Anti-vanity rules

No green. Red means action, black is normal.
Show distributions (P50/P80), not just averages.
Don’t backfill late data. If it missed the lock, it’s absent.
The ledger lives with the contract; no parallel documents.

What changes when the ledger runs

Seams stop hiding delays; the oldest item has a clock and an owner.
“Urgent” gets priced; breach receipts deter casual queue-jumping.
Rework gets traced to where it started—the seam—not dumped on teams.
Cross-team reviews shift from blame to trade-offs with receipts.

Start this week (no tooling project)

Pick your 3 highest-traffic interfaces.
Stand up a simple ledger page with the five sections.
Turn on timestamps at the seam and minimum-facts entry.
Lock a 15-minute weekly review with both owners.
Remove any field that never drives a decision.

Operating principles

Count the seam.
Price the misses.
Decide in one room.
Publish the trade-off.

Cheat sheet (print and tape up)

In/Out. Queue. Latency. Breaches. Rework.
Five numbers. One ledger. Clean seams.

Quinn’s note to operators: If an interface never changes decisions, it’s a hallway, not a seam. Treat it or close it.

Next episode (31): The Decision Shelf
How Quinn keeps decisions discoverable, time-bound, and versioned so teams don’t re-argue settled calls.

‍