On Estimations

Do not estimate. Count unitary tasks instead.

Break down work into the smallest possible tasks. By doing so, you may eliminate the need for traditional estimation altogether. Simply count the number of tasks per sprint to establish a velocity, which can predict delivery dates with approximately 10% accuracy. Run this method alongside your current practices and compare results.

This approach shifts focus from time-consuming estimation to in-depth story analysis and definition. It also eliminates the need for re-estimations as projects evolve, simplifying scope adjustments and resulting in significant time savings.

If you must estimate, use time units.

When estimation is unavoidable, use time-based units. Speak the language of management and stakeholders — since all estimations eventually convert to time anyway. Provide clear, accountable estimates that build trust and credibility.

“Points” were introduced by developers to avoid direct time commitments. Time estimates keep everyone aligned.

It doesn’t matter how you estimate.

Humans are terrible at estimating. The estimations worsen with duration or length and distance from the estimated task from us (in time and in space).

It doesn’t matter how you do it. It only matters to do it consistently. This ensures estimations can be compared over time, enabling continuous improvement based on past data.

Have a process for estimations. Follow it. Period.

Estimation is a skill that improves with structure and practice. Define a process and adhere to it. Turn it from guesswork into a disciplined task.

Pigs estimate, chickens massage.

Estimations are for pigs. No chicken interference allowed. In Agile terms, pigs (the ones doing the work) should handle the estimations. Chickens (stakeholders, managers) should not interfere.

However, before sharing estimates externally, adjust them to include buffers. Stakeholders should see conservative timelines, reducing the risk of late deliveries.

How to estimate (if you must).

• Set a secret baseline. Establish an internal overarching estimate (by the Tech Lead if it has a technical base or management if it is defined by cost or time boundaries).
• All pigs take part on it.
• Avoid pressure or lead from experienced ones. Perhaps by casting the estimations anonymously.
• Foster learning. Accept that not all estimates will be accurate and treat this as a growth opportunity.
• No chickens present. It is a private party only for pigs. Keep stakeholder influence out of the estimation room.
• Estimate only indivisible stories or tasks. This minimizes the need for future re-estimations as the project evolves.

Pig tools for estimation.

• Estimation poker.
• Developer discussions.
• Standarized task baselines. Do not reinvent the wheel.
• Have your estimation process documented as described above.

Chicken tools to adjust estimates.

• Cone of uncertainty.
• Experience modifiers (adjust for team skill levels).
• Estimation multipliers (general or individual-based, using historical data).
• Scaling time units (as data moves up the chain: hours → days → weeks → months → quarters → years).
• 3 point estimation (optimistic, most likely, pessimistic). Use PERT distributions and share the conservative figure upward.
• Burn up graphs.
• Factor in overlooked costs (holidays, sick leave, poor ticket writing, meetings, etc.).
• Project postmortems. Track past estimations and results. Adjust. Repeat. In a stable team, estimating activities will improve with time.
• An estimation sheet helps. Or other tool to facilitate inputting values and processing them.

Know your minimum estimation values.

Understand the baseline costs of a project, including organizational overhead, before starting. This knowledge is crucial for properly qualifying potential projects.

Dunning-Kruger effect.

Cognitive bias in which unskilled individuals suffer from illusory superiority, mistakenly rating their ability much higher than average. This bias is attributed to a metacognitive inability of the unskilled to recognize their ineptitude.

It’s why people often underestimate tasks they don’t fully understand, and another reason chickens should never participate in estimation sessions.

Last updated: 20 February 2025