Estimating Large Backlogs Efficiently

Estimating Large Backlogs Efficiently

When you have 50, 100, or 200+ items to estimate, planning poker one-by-one is impractical. These techniques maintain reasonable accuracy at scale.

Affinity Estimation

The fastest method for large backlogs. The entire team works simultaneously.

Process:

1. Write each story on a card (physical or digital).
2. Create columns for each scale value (1, 2, 3, 5, 8, 13, 21).
3. Team members silently place stories in columns based on gut feel. Anyone can move any card.
4. After 15 minutes, freeze movement. Discuss only items in the 13+ column.
5. Record the estimates.

Speed: 50-100 items in 30 minutes.

Bucket Sorting

Similar to affinity but with a structured pass.

Process:

1. Lay out buckets (scale values) on a table or board.
2. One person reads story titles aloud, one by one.
3. The team quickly agrees on a bucket. Use a "fist of five" (1-5 fingers) for speed.
4. If instant agreement, move on. If not, set aside for discussion.
5. Discuss set-aside items at the end.

Speed: 30-50 items in 30 minutes.

Relative Mass Valuation

For very rough sizing at the epic or initiative level.

Process:

1. Order all items from smallest to largest by effort.
2. Mark natural breakpoints ("everything above this line is Large+").
3. Assign T-shirt sizes or Fibonacci values to each segment.

Speed: 100+ items in 20 minutes.

Tips for All Methods

• Do not over-invest in accuracy. These are rough estimates for planning. You will refine as items approach the sprint.
• Use T-shirt sizes at this scale. Save Fibonacci for sprint-level refinement.
• Skip items you do not understand. Mark them for refinement rather than guessing wildly.
• Limit discussion. Set a 30-second timer for any item that sparks debate. If unresolved, set aside.

The goal at backlog scale is to answer: "How big is this relative to everything else?" Precision comes later.