Planning Poker vs. Alternative Sizing

---

The Setup

Planning Poker — invented by James Grenning, popularized by Mike Cohn — is the canonical agile estimation technique. The team reveals Fibonacci-numbered cards simultaneously after silent reflection. The reveal exposes disagreements; the conversation that follows produces shared understanding.¹

The alternatives have multiplied. Roman estimation (thumbs up/down/sideways) is faster. T-shirt sizing (S/M/L/XL) is rougher but lower-cognitive-load. Affinity sizing arranges stickies on a wall by relative size in minutes instead of hours. The debate: is planning poker worth its time cost, or have the alternatives caught up?

The Case for Planning Poker

• Forces silent reflection. The card-on-paper move makes each person commit privately before the reveal.
• The reveal surfaces disagreement. A 3 next to an 8 is the conversation starter. Without it, the loud voices drive consensus.
• Builds estimation calibration. Over time, the team's estimates align as members learn each other's mental models.
• Familiar. Most agile-trained people know the format; onboarding is trivial.

How Planning Poker Fails

• Time-expensive. Estimating twenty stories with five engineers can take an hour. The cost per estimate is high.
• Fibonacci theatre. Choosing between 5 and 8 implies a precision the team doesn't actually have.
• Encourages over-precision. The numbers feel objective. The team forgets they're rough.
• Drift to consensus too fast. Teams that have used poker for a long time often align reflexively, without the conversation the format was designed to produce.

The Alternatives

T-shirt sizing

Stories are sized S, M, L, XL (sometimes XS and XXL too). Faster than poker, lower precision, lower cognitive load. Particularly useful for stakeholder-facing conversations where Fibonacci numbers feel falsely precise.

Roman estimation

Thumbs up = yes, sideways = unsure, down = no. Used to gauge whether a story is small enough, ready enough, or risky enough — not to assign a numeric size. Very fast; surfaces disagreement quickly.

Affinity sizing

Many stories are arranged on a wall in order of relative size, without explicit scoring. Sticky notes drift left and right until the team agrees the relative ordering is right. Often produces buckets (small, medium, large) that get retrospectively scored if needed. Useful for sizing thirty stories in 30 minutes.

Bucket system

A variant of affinity sizing where stories are physically placed into buckets labeled with sizes. Allows new stories to be quickly compared to existing ones.

Magic estimation

Each team member silently places sticky notes on a scale, then the team converges. Combines silent reflection with bulk processing.

The Underlying Question

The choice depends on the team's goal for the session:

• Need calibration and conversation? Planning poker. The slow reveal is the point.
• Need to size a lot of stories quickly? Affinity or magic estimation. Bulk processing beats per-story discussion.
• Need a sanity check, not a number? Roman estimation. Fast and minimal.
• Need stakeholder-friendly rough sizing? T-shirt sizing. Feels less precise; communicates uncertainty.

The Mature Pattern

Most teams that have used poker for years eventually drift toward lighter alternatives — not because poker is bad, but because the calibration it produces has already happened. Once the team has internalized story sizing, the formal reveal becomes ceremonial. Lighter methods preserve the speed while keeping the conversation when it's needed.

Newer teams benefit from poker for exactly the opposite reason: they don't yet have shared calibration, and the explicit reveal is the mechanism that builds it.

The Common Failure

Whatever method is used, the same failure mode shows up: treating the number as the point. Whether the team estimates 8 points or "Medium," the value is in the conversation that produced the answer. A team that produces accurate numbers without conversation has missed the point as completely as one that produces inaccurate numbers with much conversation.

Coaching Tips

Use poker to build calibration.

New teams benefit from the explicit reveal. Don't skip it until the team can estimate consistently without the structure.

Try affinity for backlog sizing.

Sizing thirty stories one at a time burns out the team. Affinity sizing in batches takes a fraction of the time.

Drop Fibonacci theatre.

If the team always estimates 5 or 8, the granularity isn't real. Move to T-shirt or just count items.

Watch for fast consensus.

If poker reveals always agree, the format isn't doing work. Either the team is genuinely calibrated (drop poker) or it's anchoring on the first card revealed (fix the format).

Use Roman for refinement gates.

"Is this story small enough to take in?" thumb vote. Fast, low-cognitive-load, and surfaces disagreement immediately.

Don't conflate method with maturity.

A team using affinity sizing isn't more advanced than a team using poker. The right method depends on the situation, not the level.

Summary

The planning-poker-vs-alternatives debate is about whether the calibration value of explicit reveal is worth its time cost. For new teams, almost always yes. For mature teams, often not — the calibration has already happened, and lighter techniques preserve the conversation when it matters. The wrong move is to treat one method as universally correct. The right move is to choose the method that fits the team's current state and the session's actual purpose.

Back to The Great Agile Debates