Why performance reviews fail engineering teams
A CTO’s take on why performance reviews fail engineers, and how transparency changes everything.
Dave Garcia
Founder and Co-CEO
Feb 3, 2026
Every year, companies repeat the same ritual.
We ask managers to summarize months of engineering work into a few paragraphs and a rating. We debate whether someone is a “meets expectations” or an “exceeds expectations.” We argue about impact, ownership, attitude, leadership. And we pretend that this process is fair.
It isn’t.
After years as a CTO, I’ve come to a hard conclusion: performance reviews fail engineering teams not because managers are bad, but because the system is structurally broken.
The problem is not effort. It’s lack of transparency.
Engineering work is continuous, contextual, and cumulative. Performance reviews compress all of that into a single (or maybe two) moments in time.
What gets remembered?
The last project shipped
The loudest incidents
The most visible work
The strongest narratives
What gets lost?
Sustained delivery over time
Quiet reliability
Work without incidences
Work that reduces future risk
Contributions that don’t fit a neat story
When we compress complexity into a few infrequent checkpoints a year, we don’t measure real delivery, we end up measuring recall and perception.
Annual reviews reward storytelling, not delivery
The best engineers are not always the best self-promoters. Many of the most impactful contributions happen deep in systems, tooling, or architecture and those areas rarely shine in review meetings, specially in front of non-technical executives.
Over time, engineers learn the game:
Optimize for visibility
Avoid risky but necessary work
Bias toward short-term wins
Frame outcomes instead of reality
This is how performance reviews quietly reshape behavior—often in ways that don’t help teams scale. I’ve always been obsessed with preserving the clarity of a startup, where everything feels visible and sharp. As companies grow, that clarity tends to blur. Until now.
Calibration without evidence is just negotiation
Most organizations try to “fix” reviews with calibration sessions. In practice, these often become negotiation rooms where managers defend their people with anecdotes and selective examples.
At that point, outcomes stop reflecting performance and start reflecting advocacyhow strongly a manager believes in a promotion and how effectively they can negotiate it.
The uncomfortable truth
You cannot run a fair performance process on subjective collective memory and anecdotical datapoints alone.
That doesn’t mean reducing people to activity counts or output volume. It means grounding conversations in reality, in how work actually flows, how delivery evolves over time, and how teams truly perform.
Until we do that, performance reviews will continue to fail engineers, not because we don’t care enough, but because we’re asking a broken mechanism to do an impossible job.
I felt I had to change this, therefore I built Pensero. No more blurry sights, no more lack of transparency and feeling based performance reviews. I want to make sure everyone is in control of their career, don’t you?
Every year, companies repeat the same ritual.
We ask managers to summarize months of engineering work into a few paragraphs and a rating. We debate whether someone is a “meets expectations” or an “exceeds expectations.” We argue about impact, ownership, attitude, leadership. And we pretend that this process is fair.
It isn’t.
After years as a CTO, I’ve come to a hard conclusion: performance reviews fail engineering teams not because managers are bad, but because the system is structurally broken.
The problem is not effort. It’s lack of transparency.
Engineering work is continuous, contextual, and cumulative. Performance reviews compress all of that into a single (or maybe two) moments in time.
What gets remembered?
The last project shipped
The loudest incidents
The most visible work
The strongest narratives
What gets lost?
Sustained delivery over time
Quiet reliability
Work without incidences
Work that reduces future risk
Contributions that don’t fit a neat story
When we compress complexity into a few infrequent checkpoints a year, we don’t measure real delivery, we end up measuring recall and perception.
Annual reviews reward storytelling, not delivery
The best engineers are not always the best self-promoters. Many of the most impactful contributions happen deep in systems, tooling, or architecture and those areas rarely shine in review meetings, specially in front of non-technical executives.
Over time, engineers learn the game:
Optimize for visibility
Avoid risky but necessary work
Bias toward short-term wins
Frame outcomes instead of reality
This is how performance reviews quietly reshape behavior—often in ways that don’t help teams scale. I’ve always been obsessed with preserving the clarity of a startup, where everything feels visible and sharp. As companies grow, that clarity tends to blur. Until now.
Calibration without evidence is just negotiation
Most organizations try to “fix” reviews with calibration sessions. In practice, these often become negotiation rooms where managers defend their people with anecdotes and selective examples.
At that point, outcomes stop reflecting performance and start reflecting advocacyhow strongly a manager believes in a promotion and how effectively they can negotiate it.
The uncomfortable truth
You cannot run a fair performance process on subjective collective memory and anecdotical datapoints alone.
That doesn’t mean reducing people to activity counts or output volume. It means grounding conversations in reality, in how work actually flows, how delivery evolves over time, and how teams truly perform.
Until we do that, performance reviews will continue to fail engineers, not because we don’t care enough, but because we’re asking a broken mechanism to do an impossible job.
I felt I had to change this, therefore I built Pensero. No more blurry sights, no more lack of transparency and feeling based performance reviews. I want to make sure everyone is in control of their career, don’t you?
Every year, companies repeat the same ritual.
We ask managers to summarize months of engineering work into a few paragraphs and a rating. We debate whether someone is a “meets expectations” or an “exceeds expectations.” We argue about impact, ownership, attitude, leadership. And we pretend that this process is fair.
It isn’t.
After years as a CTO, I’ve come to a hard conclusion: performance reviews fail engineering teams not because managers are bad, but because the system is structurally broken.
The problem is not effort. It’s lack of transparency.
Engineering work is continuous, contextual, and cumulative. Performance reviews compress all of that into a single (or maybe two) moments in time.
What gets remembered?
The last project shipped
The loudest incidents
The most visible work
The strongest narratives
What gets lost?
Sustained delivery over time
Quiet reliability
Work without incidences
Work that reduces future risk
Contributions that don’t fit a neat story
When we compress complexity into a few infrequent checkpoints a year, we don’t measure real delivery, we end up measuring recall and perception.
Annual reviews reward storytelling, not delivery
The best engineers are not always the best self-promoters. Many of the most impactful contributions happen deep in systems, tooling, or architecture and those areas rarely shine in review meetings, specially in front of non-technical executives.
Over time, engineers learn the game:
Optimize for visibility
Avoid risky but necessary work
Bias toward short-term wins
Frame outcomes instead of reality
This is how performance reviews quietly reshape behavior—often in ways that don’t help teams scale. I’ve always been obsessed with preserving the clarity of a startup, where everything feels visible and sharp. As companies grow, that clarity tends to blur. Until now.
Calibration without evidence is just negotiation
Most organizations try to “fix” reviews with calibration sessions. In practice, these often become negotiation rooms where managers defend their people with anecdotes and selective examples.
At that point, outcomes stop reflecting performance and start reflecting advocacyhow strongly a manager believes in a promotion and how effectively they can negotiate it.
The uncomfortable truth
You cannot run a fair performance process on subjective collective memory and anecdotical datapoints alone.
That doesn’t mean reducing people to activity counts or output volume. It means grounding conversations in reality, in how work actually flows, how delivery evolves over time, and how teams truly perform.
Until we do that, performance reviews will continue to fail engineers, not because we don’t care enough, but because we’re asking a broken mechanism to do an impossible job.
I felt I had to change this, therefore I built Pensero. No more blurry sights, no more lack of transparency and feeling based performance reviews. I want to make sure everyone is in control of their career, don’t you?
