← Insights

Operations

Kanban in Excel, with WIP limits that bite

July 2026 · 6 min read

Draw five columns, drop your work into them, drag cards left to right as they progress. That is where most people think kanban ends, and it is why most kanban boards do nothing. A board that only shows where the work sits is a status picture. A board that limits how much work can sit in each column is a system - and the limit is the entire point.

Kanban came out of a factory, not a whiteboard. The original insight was brutally simple: a stage should not start new work while it is drowning in what it already has. Cap the work in progress, and problems that were invisible in a pile of cards suddenly have nowhere to hide.

Why WIP limits are the mechanism, not decoration

A work-in-progress limit is a number you write above a column: no more than three cards In Progress, no more than two in Review. It looks like a constraint on the team. It is actually a spotlight on the bottleneck.

Here is the logic. When Review hits its limit and stays there, the stages upstream cannot hand work forward. They are forced to stop starting and help finish. The queue that would have quietly built up behind Review - the classic place work goes to die - never forms, because the limit made the overload visible the moment it happened. Without a limit, a column can hold twelve cards and look no different from a column holding two. The pile just grows, cycle times stretch, and nobody can say why the team feels busy but nothing ships.

So the limit is not there to slow anyone down. It is there to make the constraint announce itself. A board without WIP limits is a to-do list with extra columns.

A breach the board flags for you

The trouble with WIP limits on a wall of sticky notes is that nobody counts. The limit says three; there are five cards; everyone walks past it. The limit only bites if something enforces it, and that is exactly the job a spreadsheet is good at.

Set a limit per column and let the board count cards against it on every change. The instant In Progress goes from three to four, its status turns red. No judgement, no meeting, no one eyeballing the board - the breach surfaces itself, and that red cell is your bottleneck pointing at itself. That single behaviour is the difference between a kanban board and a picture of one.

Cycle time and aging - the two numbers a board should compute

Two questions decide whether flow is healthy, and both are arithmetic a spreadsheet should never make you do by hand.

• Cycle time - how long a card takes from start to Done. Capture the created date and the done date and let the board subtract them the moment a card is closed. Track it across cards and you learn what your delivery actually costs in days, not what you hope it costs.
• Card aging - how long a card has sat in its current column right now. This is the live one. A card that has aged eleven days in Review is stuck, and stuck work is where flow quietly breaks. Aging drags the longest-waiting card to the top instead of letting it hide three rows down in a list nobody re-reads.

Aging catches problems while you can still fix them; cycle time tells you afterwards whether you did. A board that computes both is doing the thinking a wall of cards leaves to memory.

Reading the whole board at a glance

Once the board keeps score, a one-page summary falls out of it for free: how many cards total, how many done, how many in progress, how many columns are over their WIP limit, the average cycle time, and how many cards have aged past your threshold. That is a stand-up in six numbers. You stop asking "where is everything?" and start asking the only question that matters - "what is breaching, and what has aged, and why?"

A kanban board earns its place the day it can tell you something you did not already know by looking. WIP limits that flag their own breaches, and aging that surfaces stuck work, are what turn a grid of cards into that instrument.

You can run this on paper and enforce it by discipline, and some brilliant teams do. Most teams do not, the limits get ignored, the queues rebuild, and within a month the board is decoration again. The point of putting it in Excel is that the limit stops depending on anyone remembering to count.