An engineering team at a tech company was asked to speed up feature releases. They optimized for deployment velocity. Pushed more weekly updates. But soon, bugs increased and stability suffered. The company started getting more complaints.
The team had hit the target but missed the point—quality had taken a backseat to speed.
In engineering teams, metrics guide performance. But if not chosen carefully, they can create inefficiencies.
Goodhart’s Law reminds us that engineering metrics should inform decisions, not dictate them. The idea behind Goodhart’s Law was first introduced by British economist Charles Goodhart.
And leaders must balance measurement with context to drive meaningful progress.
In this post, we’ll explore the idea behind Goodhart’s Law, its impact on engineering teams, and how to use metrics effectively without falling into the trap of metric manipulation.
Let’s dive right in!
Goodhart’s Law states: “When a metric becomes a target, it ceases to be a good metric.” This concept, named after economist Charles Goodhart, highlights how when a measure becomes a target, it often loses its value as an observed statistical regularity and can distort behavior. Campbell's law is a related concept, emphasizing similar pitfalls in measurement and evaluation.
In engineering, prioritizing numbers over impact can cause issues like the following examples, illustrating the concept in action:
These are all examples of Goodhart's Law in action, where the focus on a single metric leads to unintended and sometimes negative impacts.
A classic example of this is the cobra effect, which occurred under the British government in colonial India. The government offered bounties for every dead cobra to reduce the population of venomous cobras. However, cobra breeders began to breed cobras to claim the reward, and when the bounty was withdrawn, they released the now-worthless cobras, making the problem worse. This story illustrates how public policy, when based on poorly designed metrics, can backfire due to perverse incentives.
Choosing the right words to describe measures is crucial for clarity and effective communication. It is also essential to select the right measures—what makes a good measure is its ability to reflect true goals without being easily gamed. Poorly designed metrics can have negative impacts, such as encouraging counterproductive behaviors or masking real issues.
Understanding this law helps teams set better engineering metrics that drive real improvements. It also highlights the importance of not missing the true goal—teams should focus on true goals rather than just proxies. Using multiple metrics and multiple measures can help avoid the pitfalls of over-optimizing a single number. Accurate measurements and the use of quantitative data are vital for informed decision-making.
Researchers David Manheim and Scott Garrabrant have identified four categories of Goodhart's Law: regressional, causal Goodhart, adversarial Goodhart, and Goodhart due to a third factor. These categories help explain how metrics can go wrong, whether through mistaken causality, adversarial manipulation, or unaccounted-for variables.
In business and company management, a successful strategy and solution often involve identifying essential measures that align with company, customers, and employee goals. The right metrics are identified to support continuous improvement, and company management must consider other factors and values to ensure long-term success. Metrics are often used for control purposes, but it is important to measure outcomes accurately and recognize the limitations of focusing on one metric, as this can lead to optimizing for fewer people and missing broader objectives.
Metrics help track progress, identify bottlenecks, and improve engineering efficiency. Companies often use metrics to track progress and drive decision-making, but there are risks involved if these metrics are not carefully designed.
But poorly defined KPIs can lead to unintended consequences and negative impacts:
When teams chase numbers, they optimize for the metric, not the goal. It is crucial to measure outcomes accurately to ensure that metrics reflect true progress.
Engineers might cut corners to meet deadlines, inflate ticket closures, or ship unnecessary features just to hit targets. Over time, this leads to burnout and declining quality, negatively affecting employee motivation and well-being.
Strict metric-driven cultures also stifle innovation. Developers focus on short-term wins instead of solving real problems. Values should guide the design of metrics to ensure ethical behavior and long-term success.
Teams avoid risky but impactful projects because they don’t align with predefined KPIs.
Leaders must recognize that engineering metrics are tools, not objectives. Company management plays a key role in setting effective metrics that drive improvement without causing harm. Used wisely, they guide teams toward improvement. Misused, they create a toxic environment where numbers matter more than real progress.
Metrics don’t just influence performance—they shape behavior and mindset. When poorly designed, the outcome will be the opposite of why they were brought in in the first place. The concept of metric manipulation highlights the importance of understanding the theoretical framework behind measurement and evaluation. Using the right words to describe performance is crucial, as precise language helps differentiate between various types of metrics and outcomes. When organizations focus solely on metrics, they may overlook the need to measure outcomes accurately, leading to incomplete or misleading assessments. This can result in negative impacts, such as encouraging counterproductive behaviors or undermining employee well-being. Additionally, fostering a culture where values guide behavior ensures that measurement frameworks support ethical and sustainable practices. Here are some pitfalls of metric manipulation in software engineering:
When engineers are judged solely by metrics, the pressure to perform increases. This pressure can negatively affect employee well-being, leading to stress and dissatisfaction. If a team is expected to resolve a certain number of tickets per week, developers may prioritize speed over thoughtful problem-solving.
They take on easier, low-impact tasks just to keep numbers high. Over time, this leads to burnout, disengagement, and declining morale. The negative impacts of excessive pressure include reduced creativity, increased turnover, and a decline in overall team performance. Instead of building creativity, rigid KPIs create a high-stress work environment.
A culture that emphasizes values, such as ethical behavior and long-term sustainability, can help prevent burnout and support a healthier workplace.
Metrics distort decision-making. Availability bias makes teams focus on what’s easiest to measure rather than what truly matters. This reflects the concept of metric manipulation, where the theoretical framework behind measurement is overlooked, leading to unintended consequences.
If deployment frequency is tracked but long-term stability isn’t, engineers overemphasize shipping quickly while ignoring maintenance. Using the right words to describe performance is crucial, as precise terminology helps differentiate between short-term outputs and long-term outcomes.
Similarly, the anchoring effect traps teams into chasing arbitrary targets. If management sets an unrealistic uptime goal, engineers may hide system failures or delay reporting issues to meet it. To avoid these pitfalls, organizations must measure outcomes accurately, ensuring that metrics reflect true performance and support better decision-making.
Metrics can take decision-making power away from engineers. When success is defined by rigid KPIs, developers lose the freedom to explore better solutions. This can negatively impact employee motivation, as individuals may feel their contributions are reduced to numbers rather than meaningful work.
A team judged on code commit frequency may feel pressured to push unnecessary updates instead of focusing on impactful changes. This stifles innovation and job satisfaction. Maintaining a culture where values guide decisions helps preserve autonomy and ensures that ethical and long-term considerations are prioritized over short-term metrics.
Avoiding metric manipulation starts with thoughtful leadership. Organizations need a balanced approach to measurement and a culture of transparency. Finding a solution to metric manipulation is essential for maintaining integrity and driving meaningful results.
Here’s how teams can set up a system that drives real progress without encouraging gaming:
Leaders play a crucial role in defining metrics that align with business goals. Instead of just assigning numbers, they must communicate the purpose behind them.
The right metrics are identified by analyzing which measures most directly influence desired business outcomes. Among these, some metrics are essential for driving success and avoiding common pitfalls.
For example, if an engineering team is measured on uptime, they should understand it’s not just about hitting a number—it’s about ensuring a seamless user experience.
When teams understand why a metric matters, they focus on improving outcomes rather than just meeting a target.
Numbers alone don’t tell the full story. Blending quantitative and qualitative metrics ensures a more holistic approach. Incorporating quantitative data alongside qualitative insights provides a comprehensive understanding of performance.
Instead of only tracking deployment speed, consider code quality, customer feedback, and post-release stability. Using multiple metrics and multiple measures to evaluate success helps avoid the pitfalls of relying on a single indicator.
For example, a team measured only on monthly issue cycle time may rush to close smaller tickets faster, creating an illusion of efficiency. Accurate measurements are essential to ensure that performance is assessed correctly.
But comparing quarterly performance trends instead of month-to-month fluctuations provides a more realistic picture.
If issue resolution speed drops one month but leads to fewer reopened tickets in the following quarter, it’s a sign that higher-quality fixes are being implemented.
This approach prevents engineers from cutting corners to meet short-term targets.
Silos breed metric manipulation. Cross-functional collaboration helps teams stay focused on impact rather than isolated KPIs. Embedding values into the organizational culture further promotes transparency, ensuring that ethical considerations guide both decision-making and measurement.
There are project management tools available that can facilitate transparency by ensuring progress is measured holistically across teams.
Encouraging team-based goals instead of individual metrics also prevents engineers from prioritizing personal numbers over collective success. This approach positively impacts employee motivation, as individuals feel their contributions are recognized within the broader context of team achievement.
When teams work together toward meaningful objectives, there’s less temptation to game the system.
Static metrics become stale over time. Teams either get too comfortable optimizing for them or find ways to manipulate them.
Rotating key performance indicators every few months keeps teams engaged and discourages short-term gaming. It is essential to periodically review and update key measures to ensure they remain relevant and effective.
For example, a team initially measured on deployment speed might later be evaluated on post-release defect rates. This shifts focus to sustainable quality rather than just frequency.
Leaders should evaluate long-term trends rather than short-term fluctuations. If error rates spike briefly after a new rollout, that doesn’t mean the team is failing—it might indicate growing pains from scaling.
Accurate measurements taken consistently over time are essential for identifying meaningful trends and avoiding misinterpretation of short-term changes. Looking at patterns over time provides a more accurate picture of progress and reduces the pressure to manipulate short-term results.
By designing a thoughtful metric system, building transparency, and emphasizing long-term improvement, teams can use metrics as a tool for growth rather than a rigid scoreboard.
A leading SaaS company, known for its data-driven approach to metrics, wanted to improve incident response efficiency, so they set a key metric: Mean Time to Resolution (MTTR). The goal was to drive faster fixes and reduce downtime by accurately measuring outcomes. However, this well-intentioned target led to unintended behavior.
To keep MTTR low, engineers started prioritizing quick fixes over thorough solutions. Instead of addressing the root causes of outages, they applied temporary patches that resolved incidents on paper but led to recurring failures. Additionally, some incidents were reclassified or delayed in reporting to avoid negatively impacting the metric.
Recognizing the issue, leadership revised their approach. They introduced a composite measurement that combined MTTR with recurrence rates and post-mortem depth—incentivizing sustainable fixes instead of quick, superficial resolutions. The right metrics were identified by analyzing which inputs and outputs best reflected true system health. They also encouraged engineers to document long-term improvements rather than just resolving incidents reactively.
This shift led to fewer repeat incidents, a stronger culture of learning from failures, and ultimately, a more reliable system that improved customer satisfaction, rather than just an artificially improved MTTR.
To prevent MTTR from being gamed, the company deployed a software intelligence platform that provided a comprehensive solution for deeper insights beyond just resolution speed. It introduced a set of complementary metrics to ensure long-term reliability rather than just fast fixes.
Key metrics that helped balance MTTR:
Using multiple metrics to evaluate performance helped avoid the pitfalls of relying on a single measurement and provided a more accurate picture of system health.
By monitoring these additional metrics, leadership ensured that engineering teams prioritized quality and stability alongside speed. Accurate measurements were essential for tracking progress and identifying areas for improvement. The software intelligence tool provided real-time insights, automated anomaly detection, and historical trend analysis, helping the company move from a reactive to a proactive incident management strategy.
Company management played a crucial role in implementing these tools and fostering a culture that values comprehensive measurement and continuous improvement.
As a result, they saw:
✅ 50% reduction in repeat incidents within six months.
✅ Improved root cause resolution, leading to fewer emergency fixes.
✅ Healthier team workflows, reducing stress from unrealistic MTTR targets.
No single metric should dictate engineering success. Software intelligence tools provide a holistic view of system health, helping teams focus on real improvements instead of gaming the numbers. By leveraging multi-metric insights, engineering leaders can build resilient, high-performing teams that balance speed with reliability.
Engineering metrics should guide teams, not control them. When used correctly, they help track progress and improve efficiency. But when misused, they encourage manipulation, stress, and short-term thinking.
Striking the right balance between numbers and why these numbers are being monitored ensures teams focus on real impact. Otherwise, employees are bound to find ways to game the system.
For tech managers and CTOs, the key lies in finding hidden insights beyond surface-level numbers. This is where Typo comes in. With AI-powered SDLC insights, Typo helps you monitor efficiency, detect bottlenecks, and optimize development workflows—all while ensuring you ship faster without compromising quality.
Take control of your engineering metrics.
Implement engineering metrics, gather SDLC insights & set continuous improvement goals in your dev teams.
Try Live Demo
