Why we still do performance reviews at Cabify

Performance evaluations. Just the phrase alone can trigger anxiety in employees and managers alike. Critics argue they’re broken and frustrating, while defenders say they’re essential for accountability and growth. The truth? Traditional performance reviews are broken, but scrapping them entirely isn’t the answer.

The good news? Some engineering organizations have found ways to extract real value from performance reviews without the usual pain. Instead of throwing out the whole system, they’ve redesigned it around transparency, collaboration, and better data. At Cabify, we’re one of those organizations, and we’ll show you how we did it.

The problem with traditional performance reviews

Let’s take a step back, assume we don’t just hate performance reviews for irrational reasons, and figure out their most common problems. To fix something, we need to know what is broken.

Most performance evaluation systems fail because they commit at least one of these mistakes:

• Excessive focus on the past, ignoring growth. People get judged on what they did, rather than how they can improve. No one loves being assigned a (performance) label, but most people want to develop their skills and grow professionally.
• Lack of transparency. Many reviews feel like black boxes, where employees don’t know how decisions are made.
• Too much subjectivity. Bias creeps in when a single manager’s opinion dictates someone’s future.
• Rewarding individuals over teams. Employees may prioritize personal achievements instead of collaborating effectively.

The result? People game the system, managers dread the process, and nobody really improves.

Many people, when asked individually, would prefer not going through all of this. However, as a collective, most engineering organizations keep running performance reviews. There must be a reason for that!

Why performance reviews are still necessary

If performance reviews are so broken, why do most engineering organizations keep running them? Because when done right, they solve real problems that informal feedback can’t address at scale.

As companies grow beyond 50-100 people, informal feedback and management-by-walking-around breaks down. Leadership needs a systematic way to understand how people are performing across teams and make fair decisions about development, promotions, and compensation. Performance reviews can provide that clarity and consistency.

When every employee has a performance label, decisions about salaries, training budgets, promotions, or (worst case) layoffs can be based on that label. These labels also let us compare performance across different teams and functions.

Putting people into boxes might feel dehumanizing, especially for those who remember smaller companies where you knew everyone personally. But reducing information as it gets rolled up is a must to avoid information overload as the organization grows beyond a single management layer.

The process can help your company or just create busywork. Done wrong, it’s a perfect example of “garbage in, garbage out”: bad data leads to wrong labeling, which leads to bad decisions. It becomes all smoke and mirrors.

But lack of information can be just as harmful as misinformation. The solution isn’t to eliminate performance reviews. It’s fixing them by cutting the busywork and focusing on results.

How Cabify does it differently

So, how do you fix something that’s fundamentally flawed but still necessary? At Cabify, we’ve learned that the key is addressing the root causes: lack of context, subjectivity, and poor data quality. Instead of throwing out the whole system, we’ve redesigned it around transparency and collaboration.

• We add a squad calibration exercise that sets the context before discussing individual performance. Tech leaders and their stakeholders meet to discuss squad performance. They focus on two key areas: choosing the right problems and delivering solutions. We share these results with squad leaders before they review their team members.
• We have a clear career path to base performance reviews on. This makes things more objective when evaluating the past and helps us look forward. It is designed to prevent career development from becoming a checklist.
• We make Talent Mapping (our process for rating performance and potential) a group exercise. Managers work in groups and share their initial assessment (performance + potential) and reasoning for each person on their squad. Other managers can challenge these assessments. This peer review reduces errors and helps us align on standards.
• We keep different levels of detail for different purposes. Performance reviews are a snapshot at one point in time. Multiple other mechanisms (like 360 feedback, 1:1s and team retrospectives) are required to gather detailed information in short feedback loops. Without those, you can’t create a useful snapshot (“garbage in, garbage out”).

Let’s dive deeper into each of these approaches.

Squad calibrations

Squad calibration is our term for evaluating team performance before individual performance. We originally created this process to prevent wildly different opinions across disciplines. We’ve reduced this risk through regular feedback, but we found the calibration process so valuable that we kept it anyway.

We fail and succeed together. Building software is a social activity that goes beyond Engineering. That’s why we start by reviewing how tech squads (composed of multiple disciplines such as engineering, product, design, data…) performed as seen by their leaders and stakeholders.

For example, a squad might score high on picking the right problems because they identified a critical user pain point, but score lower on delivering solutions because they struggled with technical debt. This context helps engineering managers understand what individual performance issues might stem from team challenges.

Engineering managers use this calibration as context when they then evaluate individuals. It would be very unusual for a high-performing team to be made up of low-performing individuals, and vice versa.

Career paths

We use a career path that’s common across the whole company. It’s built around our core company values plus discipline-specific values (in Engineering we’ve chosen Nimbleness and Excellence). This mix of shared and specific values helps push the whole company in the same direction. At the same time, it recognizes that different disciplines work differently. Each value describes specific behaviors that show everyone what’s expected at their current level and what they need to grow.

Our latest career path for Engineers was created by a working group of Staff engineers and Senior Managers. This approach reflects actual day-to-day work, not just leadership’s vision of the work.

By design, career path behaviors are broad. This allows different ways to succeed and prevents people from just checking boxes without actually contributing to company success. Which brings us to the next point…

Making talent mapping collaborative

Talent Mapping is how we rate each person’s current performance and future potential. We know that career paths can’t cover every detail, and there will always be room for interpretation. More than that, we use this flexibility to make space for different profiles (like rock stars and superstars).

This means we need a way to align our interpretations. Instead of adding more documentation and process, we just get together on a video call. We propose mappings for each person on our teams, ask for clarifications, step back to look at the big picture, make adjustments… and we’re done.

It’s not complex, but with close to 200 people in Engineering, this takes time (about 15 hours, not counting prep work). Not every manager joins every meeting, of course. We split the sessions by similar levels, which helps with cross-engineering alignment.

We hear this is one big difference from other companies. Some encourage performance mapping to be done alone, while others create fierce competition between managers. While our process could potentially backfire and turn into competition, in practice it hasn’t.

Matching detail to purpose

The performance review gives us a useful big-picture view of the company (a zoomed-out, low-detail tool). It’s great for avoiding information overload while still allowing company-wide decisions. But this works by removing most of the details. It’s a good tool for seeing performance at a point in time, but it’s not the right tool for managing performance.

This is why the zoomed-in, detailed interactions within teams stay critical. Without them, there’s nothing to take a snapshot of. Before each performance cycle, we gather 360 feedback from peers, stakeholders, and direct reports to get multiple perspectives on each person’s contributions. We combine this with ongoing interactions like 1:1 meetings, Development Plans, Objective Plans, dailies, team and project retrospectives…

Improving performance reviews at your company

You don’t need to copy our entire system to improve your performance reviews. Here are three changes you can start with:

Start with team context. Before evaluating individuals, spend time with other managers and stakeholders discussing how each team performed. Ask: “What challenges did this team face? What did they achieve?”. Your stakeholders already have opinions on your teams’ performance and making them explicit helps you take action.

Make assessments collaborative. Instead of managers working in isolation, try reviewing a few people together. Challenge each other’s reasoning. You’ll be surprised how often “high performer because they shipped a lot” becomes “average performer who prioritized velocity over impact”. You might also discover that different teams use different criteria when they shouldn’t. When this happens, you’ll need to work on converging those standards.

Keep it simple. Don’t try to fix everything at once. Pick one element (like team context or collaborative assessment) and run it for one cycle. Learn what works in your culture before adding more complexity.

Final thoughts: The right way to fix performance reviews

Line Managers, Heads, VPs, and C-Level executives need different zoom levels to do their jobs. The People department needs standardized information from all areas to plan company-wide improvements, from training to compensation strategies. Individual contributors have little say, yet they feel like they’re under a microscope.

It’s impossible to have a one-size-fits-all performance process that makes everyone happy. So we’re unlikely to fix performance reviews and make them engaging and joyful. But by knowing what we want from them, we can design them to be more valuable and less painful than usual. At Cabify, we’re consciously trying to do this, one cycle at a time.