The fallacy of estimates and effective agile alternatives

By Josep Vives (ERNI Spain)

A professional approach to business agility

In the dynamic world of business agility, estimating the duration of a product, task or process is a commonly used practice. However, an increasing number of voices are speaking out against this practice, arguing that spending time on such estimates is more detrimental than beneficial.

In this article, we explore why estimates fail and consider alternatives such as the #NoEstimates movement and flow metrics in Kanban. We also examine how Monte Carlo simulations, when adapted to agile teams, can offer more useful and objective tools based on real data.

Are estimates a waste of time?

Many teams spend a significant amount of time and patience estimating. How long will a task take? How long it will be before the deliverable is in the customer’s hands? These two questions, while related, are not the same.

It was thought that estimating how much work needed to be done would answer these “million-dollar” questions – basically, predicting the future. However, these predictions often turn out to be inaccurate, unreliable and frustrating.

Given the difficulty of accurately predicting the number of hours required for a task, relative estimation was introduced (along with the nightmare of story points). This approach involves comparing a task’s size, complexity or uncertainty to a reference task assigned a value, such as “1 SP”. If the task seems twice as big, it is assigned “2 SP”; if it appears five times larger, it is assigned “5 SP”, and so on. While this method helps provide a general sense of the required work and adapts well to changing requirements, it is flawed.

The real problem arises when these SPs are converted into days or hours to communicate with stakeholders – figures that rarely reflect reality.

No matter how good we are at comparisons, uncertainties in requirements, cognitive biases, deadline pressures and the lack of reliable historical data – combined with factors beyond our control – often lead agile teams to focus solely on meeting deadlines. This results in frustration as well as the inefficient use of valuable resources, commonly referred to as waste.

An estimate is not a commitment, and a forecast is not a plan

Before embarking on any planning or estimation process – breaking down stories, assigning story points or creating tasks – it’s essential to focus on the larger issue at hand. Agile’s core principle is the ability to progress with incomplete information while rapidly adapting as better information becomes available. However, a lack of organisational understanding of agile principles remains a significant challenge.

The NoEstimates movement

In response to these challenges, the controversial NoEstimates movement emerged, advocating a radically different approach. This philosophy encourages teams to move away from focusing on estimates and instead adopt more agile and adaptive practices.

The core tenets of NoEstimates include:

• Focus on value: Rather than predicting duration, teams should prioritise continuous and rapid delivery of value.
• Work in small increments: Breaking work into manageable pieces allows for constant feedback and course correction as needed.
• Flow metrics: Tools and techniques like Kanban offer clear visualisation of workflow, helping identify bottlenecks and drive continuous improvement.

The movement strongly emphasises two values from the Agile Manifesto:

1. Customer collaboration over contract negotiation
2. Responding to change over following a plan

To implement these, NoEstimates suggests four steps for teams:

1. Select the most prioritised feature (ideally the one delivering the most value).
2. Break the work into risk-neutral chunks that won’t jeopardise project goals if incomplete.
3. Develop each chunk according to the Definition of Done.
4. Iterate and refactor.

Flow metrics in Kanban

Kanban, a visual work management methodology, relies on principles that help teams manage workflow more efficiently. Key metrics include:

• Lead Time: Total time from starting a process to completion, measuring workflow efficiency.
• Cycle Time: Time taken for a task to move from “In Progress” to “Done”, helping identify bottlenecks.
• Throughput: The amount of work completed over a given period, indicating the team’s capacity to deliver value.
• Work In Progress (WIP): The number of items started but not completed, typically seen in the “In Progress” column.
• Work Item Age: Tracks how long an item has been in the “In Progress” column, detecting delays and blockages.

These metrics enable teams to measure, analyse and improve workflow, enhancing value delivery to customers.

Monte Carlo simulations

Monte Carlo simulations, originally developed in the 1940s for neutron radiation protection, have become valuable in agile development. This statistical method simulates multiple scenarios using variables to assess risks and predict outcomes.

Benefits for agile teams:

• Data-driven forecasting: Uses team historical data for realistic predictions.
• Risk analysis: Identifies and quantifies risks to prepare for potential issues.

Key questions answered:

1. How many items can be completed in a given timeframe? Percentiles illustrate variability. For example, if the 85th percentile predicts 156 items, there’s an 85% chance of completing at least 156 items within the timeframe, leaving a 15% risk of falling short.
2. How long will it take to finish specific work? Based on cycle and lead times, Monte Carlo simulations can predict timelines with a given certainty. For instance, there might be an 85% certainty of completing a task in 23 days or within a shorter timeline with higher risk.

Teams already using tools like Cycle Time Scatterplots or Throughput Histograms gain insights into their capacity and reasonable future commitments, moving beyond simplistic calculations based on headcount or velocity.

Overcoming resistance to change

Resistance to abandoning traditional estimation methods stems from the comfort and predictability they appear to provide. Fear of failure and pressure to meet expectations reinforce adherence to familiar methods. However, simplifying processes often brings significant benefits.

For years, mathematics and statistics have helped large organisations manage business risks. Businesses can make smarter decisions by thinking probabilistically, focusing on relevant data, and continuously refining workflows rather than relying on intuition or “gut feelings”.

By shifting from traditional estimates to data-driven approaches, agile teams can manage work more effectively, minimise waste, and focus on delivering continuous value.

Would you like to see how agile works at scale? Discover how one of the world’s 100 largest tech companies applies agile principles to drive efficiency and innovation in this article: Agile in one of the 100 largest technology companies in the world – Context and impact – ERNI.