Using Story Points in Kanban (Not Just Scrum)

Story Points in Kanban

Story points are often associated with Scrum sprints, but they can be valuable in Kanban workflows too -- with some adaptation.

Do You Even Need Points in Kanban?

Kanban focuses on flow metrics: cycle time, lead time, throughput, and WIP limits. Many Kanban teams skip estimation entirely and rely on these metrics for forecasting.

When points add value in Kanban:

• Planning future work capacity ("Can we finish these 50 items by Q3?")
• Communicating relative size to stakeholders
• Identifying items that should be split before entering the board
• Teams transitioning from Scrum who want to keep estimation

When to skip points:

• All items are roughly the same size (common in maintenance teams)
• The team has enough historical throughput data for Monte Carlo forecasting
• Estimation sessions are a bottleneck in your flow

Adapting Estimation for Kanban

No sprint cadence = no sprint planning. Instead:

1. Estimate during refinement. As items enter the backlog, the team estimates them individually or in small batches.
2. Use WIP limits as a forcing function. If a 13-point item would monopolize the board, it must be split.
3. Track both points and count. Count-based throughput is simpler, but point-based throughput helps when item sizes vary.

Forecasting Without Sprints

In Scrum, velocity (points/sprint) drives forecasting. In Kanban, use:

Points-based throughput:

Last 4 weeks:
  Week 1: 22 points delivered
  Week 2: 18 points delivered
  Week 3: 25 points delivered
  Week 4: 20 points delivered

  Average: 21.25 points/week
  Remaining backlog: 150 points
  Forecast: ~7 weeks

Count-based throughput (alternative):

Last 4 weeks:
  Week 1: 8 items completed
  Week 2: 6 items completed
  Week 3: 9 items completed
  Week 4: 7 items completed

  Average: 7.5 items/week
  Remaining backlog: 45 items
  Forecast: ~6 weeks

Monte Carlo Simulation

For probabilistic forecasting, run a Monte Carlo simulation using historical throughput data. This gives you confidence intervals instead of a single date:

"There is an 85% chance we will finish the backlog
 within 6 to 9 weeks."

This is more honest than "we'll be done in 7 weeks" and accounts for natural variation in throughput.