Story Points vs. NoEstimates

---

The Two Camps

Few debates in agile attract more religious fervor than this one. On one side: story points, the relative-sizing tool popularized by Mike Cohn and James Grenning in the early 2000s, used by most Scrum teams as the unit of estimation for sprint planning and velocity tracking.¹ On the other: NoEstimates, the movement led by Vasco Duarte, Woody Zuill, and others, which argues that estimation is a wasteful ritual and that throughput-based forecasting produces better predictions with less effort.²

Both camps are correct about something, and both are wrong about something. The point of this debate is not to declare a winner but to make the trade-offs visible enough to choose deliberately.

The Case for Story Points

Story points are not really about predicting effort. Their original justification was different and more interesting: relative sizing forces a conversation. When the team has to decide whether a story is a 3 or a 5, the discussion exposes assumptions, surfaces unknowns, and aligns understanding before development begins.

Cohn's original argument: story points compress a multidimensional reality (complexity, uncertainty, effort, risk) into a single relative scale, freeing the team from the false precision of time estimates. The unit is intentionally abstract — points cannot be converted to hours, and that is the point. They measure size, not duration.

The downstream payoff: a team's velocity (points per sprint) becomes a usable input for release planning, and the act of estimation itself produces shared understanding as a side effect.

The Case for NoEstimates

The NoEstimates argument is empirical and structural. Empirically, studies show that simple throughput-based forecasting — counting completed items per week and projecting forward — produces predictions at least as accurate as point-based velocity, and often more so.³ Structurally, the time spent on estimation is largely waste: it does not improve the work, the conversation it generates can happen without numerical scoring, and the resulting numbers tend to be misused by managers as commitments.

NoEstimates proponents also point out a second-order effect: when teams estimate, they tend to break stories down to the size that fits the estimation game (e.g., a 5-point story). When teams forecast on throughput, they tend to break stories down to whatever size flows well. The latter tends to produce smaller, healthier stories.

Where Each Approach Breaks

Story points break when:

• The team becomes velocity-obsessed and gameable. Points get inflated to look productive.
• Stakeholders ask "how many points is that?" as a substitute for "is it valuable?"
• Sprint commitments become contracts. The team starts padding estimates defensively.
• Cross-team comparisons begin. Point velocities are not comparable across teams; treating them as if they are produces dysfunction.

NoEstimates breaks when:

• The team has not yet built the discipline of consistent story sizing. Throughput forecasting assumes stories are roughly similar in size — which only happens through deliberate slicing practice.
• Stakeholders need long-range forecasts and the team has only weeks of historical data.
• The work is genuinely heterogeneous (research, novel features, hardware) and historical throughput is not predictive.
• The team uses NoEstimates as an excuse to skip the conversation that estimation provoked. The numbers were never the point; the conversation was.

The Hidden Agreement

Both camps actually agree on more than they realize. Both believe:

• Small stories beat large stories.
• Shared understanding produced by conversation matters more than the resulting number.
• Velocity-as-target is dysfunctional.
• Long-range "exact" predictions are dishonest regardless of method.
• Forecasting should serve decisions, not perform control.

The actual disagreement is narrower than the rhetoric suggests: do we need a numerical sizing step to produce the conversation, or does small-story discipline plus throughput suffice?

A Pragmatic Stance

Most healthy teams end up somewhere in the middle. A working approach:

• Use coarse sizing (S/M/L, or simple counts like 1, 2, 3) during refinement to expose differences in understanding.
• Forecast based on throughput, not point-velocity, for release planning.
• Aim for stories that are mostly the same size — the strongest predictor of either method's accuracy.
• Drop fine-grained Fibonacci scoring (3 vs. 5 vs. 8) once the team is consistently slicing well. The conversation, not the precision, was always the value.

What This Debate Is Actually About

Underneath the noise, this debate is a proxy for a deeper question: does the team need a heavy ritual to produce shared understanding, or has it developed enough trust and discipline to produce that understanding without one? Story points are training wheels for the conversation. NoEstimates assumes the team has graduated from the training wheels. Both are right at the appropriate point in a team's maturity. The mistake is treating either as a permanent identity.

Coaching Tips

Listen for "how many points?"

When stakeholders ask about points instead of value, the number has become the point of the practice. That's the smell, not the team's choice of method.

Measure story-size consistency.

Whichever method the team uses, the underlying question is whether stories are roughly the same size. That's what makes forecasting work.

Forecast in ranges.

"7 to 12 stories" beats "9 points" for honesty. Both methods produce ranges if used correctly.

Try the conversation without numbers.

Run one refinement using qualitative sizing only ("small / medium / large"). See if the conversation still produces shared understanding. Often it does.

Don't make it an identity.

"We don't do points" or "we always do points" both freeze the practice. The right approach changes with team maturity.

Track forecast accuracy.

Pick a method, predict, measure. The empirical question — does this method actually predict? — is more useful than the philosophical one.

Summary

The story points vs. NoEstimates debate generates more heat than its underlying disagreement deserves. Both camps want small stories, useful forecasts, and an end to false precision. They differ on whether numerical sizing is the engine of the necessary conversation or an obsolete ritual masquerading as one. The honest answer depends on the team's current discipline — and the team's current discipline can change.

Back to The Great Agile Debates