I Experimented with ChatGPT to Improve My Annual Performance Review. Here's What I Learned.
Can ChatGPT make a sometimes frustrating process easier?
Created on February 29|Last edited on January 12
Comment
Introduction
Annual performance reviews can be tricky to write. Logically summarizing an entire year’s work is no small task and it can be hard to know where to begin. Striking a tone that feels both professional and confident but not boastful is a tough needle to thread and, for me personally, I’m frequently my own worst critic. I struggle with owning my accomplishments and strengths and generally find this far more difficult than identifying my weaknesses.
There’s a special irony here. I’ve been in People Operations for nearly two decades and have either participated in or set the strategy for annual performance views in every one of those years. This year, I wrote my performance review and ran into the same challenges I have in the past. But this year, I’m working at an ML tools company with ready access to ChatGPT.
Which got me thinking: could ChatGPT make this process better? Could it improve my actual performance review?
I decided to find out. And what I discovered was that yes, ChatGPT could in fact make my review better. With just a handful of prompts, I ended up replacing several aspects of the review I’d written myself with content ChatGPT provided. It needed a little clean up—and we’ll get into exactly what needed to be done further down in the piece—but I was anticipating that. What I didn’t anticipate was how good the output would be without all that much effort on my end.
Let’s get into the process:
The first attempt
Since I’d taken a first pass at writing my review the old fashioned way, I had a solid idea about what I wanted to highlight. I’m not a seasoned pro at prompt engineering so I anticipated the first attempt may miss the mark. I was right.
Here’s the prompt I used:
You are writing an annual performance review for a Chief People Officer at a 200 person machine learning company. They’ve been with the company for about six months. Their impact during that time has been as follows:
1. revamping the annual performance process to focus on feedback, growth and impact
2. finding a new office space to support in person collaboration
3. Completing benefits open enrollment to align benefits to the market
4. Making progress on running a data driven People Team with the first people quarterly business review
And here’s the rather verbose output of that prompt:
[Mariah] has demonstrated exceptional leadership and strategic vision in spearheading various key initiatives aimed at fostering a thriving work environment for our employees. She has played a pivotal role in revamping our annual performance process, finding a new office space to support in-person collaboration, and completing benefits open enrollment to align benefits with market standards.
Key Achievements:
Revamping Annual Performance Process: [Mariah] led the effort to revamp our annual performance process, focusing on enhancing transparency, accountability, and employee development. Through her strategic insights and collaborative approach, she successfully implemented a performance evaluation system that promotes continuous feedback, goal alignment, and professional growth opportunities.
Finding New Office Space: Recognizing the importance of fostering in-person collaboration and providing a conducive work environment, [Mariah] spearheaded the search for a new office space. She evaluated various options, considering factors such as location, space requirements, and amenities. Her efforts culminated in securing a new office space that aligns with our company's growth trajectory and fosters a sense of community among our employees.
Completing Benefits Open Enrollment: Understanding the significance of offering competitive benefits to attract and retain top talent, [Mariah] took charge of completing benefits open enrollment. She collaborated with relevant stakeholders to review and update our benefits package, ensuring that it remains competitive and aligned with market standards. Her meticulous attention to detail and comprehensive approach ensured a smooth and efficient open enrollment process for our employees.
Areas for Development:
While [Mariah] has made remarkable contributions in her role as Chief People Officer, there are areas where further development and focus could enhance her effectiveness:
- Employee Engagement Initiatives: Exploring additional initiatives to further enhance employee engagement and promote a positive work culture.
- Talent Development Programs: Developing and implementing talent development programs that foster continuous learning and career advancement opportunities for our employees.
Conclusion:
In conclusion, [Mariah] has made significant contributions to our organization in her capacity as Chief People Officer over the past six months. Her leadership, strategic acumen, and dedication to enhancing our people operations initiatives have been instrumental in driving positive change and fostering a supportive work environment. I have full confidence that she will continue to lead with excellence and contribute to the success and growth of our company in the years to come.
So what worked here? ChatGPT logically summarized my impact and accomplishments, and produced a relevant introduction and conclusion that made my achievements clear. It even anticipated development areas based on my role that were not a part of the initial prompt. To me, this is both impressive and one of the most powerful aspects of using ChatGPT for your performance review.
What didn’t work for me was the writing itself. The tone is overly formal and the review is too long. ChatGPT can occasionally pad out sections needlessly, adding unnecessary adjectives or restating earlier comments in vaguely different language.
I decided to alter my prompt slightly and try again.
The second attempt
Looking back, I didn’t change all that much here but the outcome differed in some really interesting ways. My prompt for my second attempt had the same achievements but, crucially, I asked ChatGPT to write a self-review as opposed to an annual performance review. The prompt, verbatim, with bolding around changes:
You are writing a self review as a Chief People Officer at a 200 person machine learning company. You’ve been with the company for about six months. Their impact during that time has been as follows:
1. revamping the annual performance process to focus on feedback, growth and impact
2. finding a new office space to support in person collaboration
3. Completing benefits open enrollment to align benefits to the market
4. Making progress on running a data driven People Team with the first people quarterly business review.
And the output:
As I reflect on my first six months as Chief People Officer, I am proud of the progress we've made in several key areas. Here's a rundown of the impact I've had:
- Revamping the Annual Performance Process: One of my primary objectives upon joining the company was to revamp our annual performance process. I recognized the need to shift the focus towards feedback, growth, and impact. By implementing new systems and processes, we've created a more transparent and development-oriented performance evaluation system. Employees now have clearer expectations and opportunities for growth, leading to increased engagement and motivation across the board.
- Finding a New Office Space: Another significant accomplishment during this period was finding a new office space to support in-person collaboration. Understanding the importance of a conducive work environment, I spearheaded the search for a space that would foster creativity and teamwork. The successful relocation to our new office has already proven to enhance collaboration and employee satisfaction.
- Completing Benefits Open Enrollment: Aligning our benefits to market standards was a crucial task, and I'm pleased to have successfully completed benefits open enrollment. By conducting a thorough review of our benefits package, we ensured that our offerings are competitive and meet the needs of our diverse workforce. This initiative has strengthened our employer brand and contributed to employee retention and satisfaction.
- Making Progress on Running a Data-Driven People Team: Transitioning towards a data-driven approach within our People Team has been an ongoing goal. I'm pleased to report that we've made significant progress in this area, culminating in the successful implementation of our first people quarterly business review. By leveraging data analytics, we've gained valuable insights into employee performance, engagement, and retention. This data-driven approach will continue to guide our strategic decisions and drive organizational success.
Overall, I am pleased with the impact I've had during my first six months as Chief People Officer. However, there is always room for improvement, and I am committed to continuing to drive positive change and contribute to the success of our company in the months and years ahead.
This feels a lot better to me, personally. The tone is confident without some of the baroque vocabulary I got the first time around. That reduced formality is a lot more aligned with how I write and it was truly interesting that such a small change to the input could have such broad, positive effects on the output. I thought overall ChatGPT did a nice job linking my accomplishments to actual impact but it also didn’t surface any development areas (despite mentioning continued development in the conclusion).
The third attempt
I decided to handle that omission in my next prompt. Here, I kept the “self review” language that led to better tone and leaner copy but also asked specifically for development areas. The third prompt, with what I changed bolded for clarity:
You are the Chief People Officer at Weights and Biases and you are writing your annual self assessment. You've been with the company for six months. Your impact during that time has been as follows:
1. revamping the annual performance process to focus on feedback, growth and impact
2. finding a new office space to support in person collaboration
3. Completing benefits open enrollment to align benefits to the market
4. Making progress on running a data driven People Team with the first people quarterly business review.
Areas of opportunity are onboarding and manager training.
And the output:
Reflecting on my first six months as Chief People Officer at Weights and Biases, I am proud of the accomplishments we've achieved together as a team. However, I also recognize areas where there is room for improvement. Here's a breakdown of my self-assessment:
Areas of Impact:
1. Strategic Initiatives: I successfully spearheaded several strategic initiatives, including revamping our annual performance process, finding a new office space, and aligning benefits to the market. These efforts have positively impacted our company culture and employee satisfaction.
2. Data-Driven Approach: Implementing the first people quarterly business review and making strides towards running a data-driven People Team have been significant achievements. Leveraging data analytics has allowed us to gain valuable insights into our workforce and make informed decisions.
3. Leadership and Collaboration: My ability to lead cross-functional teams and foster collaboration has been instrumental in driving organizational change. Building strong relationships with stakeholders across the company has enabled us to achieve our goals effectively.
Areas for Continued Development:
1. Onboarding: While we have made strides in various areas of people operations, there is an opportunity to improve our onboarding process. Enhancing our onboarding program will ensure that new hires feel welcomed, supported, and equipped for success from day one.
2. Manager Training: Investing in manager training is essential for fostering strong leadership and employee development. Providing managers with the necessary tools and resources to effectively lead their teams will drive performance and engagement throughout the organization.
Moving forward, I am committed to addressing these areas of opportunity and continuing to drive positive change within our organization. By focusing on improving our onboarding process and investing in manager training, we can further enhance our company culture and support the growth and development of our team members.
Again, just that one sentence dramatically changed the LLM output. This prompt yielded different summarized areas of strength that align well with some of the core responsibilities of a Chief People Officer. It proactively connected my work to its impact. And it did an admirable job with the areas for improvement with minimal information. My one complaint would be that the output here is a little too high-level to really capture the day-to-day impact I’ve seen, but again the tone was a lot more conversational and, I think, appropriate for a self-assessment.
What I learned using ChatGPT for my self review
Here are three high-level takeaways from my experiment:
Small changes make a big difference
Swapping “annual performance review” with “self review” transformed ChatGPT’s output. Where the original prompt gave us more florid language, the “self review” prompt was far less verbose. Adding a single line about “areas of opportunity” worked out quite well, all things considered. ChatGPT obviously doesn’t know as much about my year as I do, so of course I had to add relevant metrics and program particulars to highlight the most important work I’ve done over the past six months. You can try adding that to a prompt or just adding it to the output.
A cure for writer’s block
I mentioned this in the introduction, but I can find it hard to start assessments. ChatGPT does not have this problem. In fact, I think it’s a great tool for simply getting a draft going and not facing the problem of staring at a blank page for hours. It can organize your thoughts in ways you may not have thought about and offer turns of phrase that resonate with you.
A partner more than a replacement
To answer the question I asked in the title: yes, ChatGPT can write performance reviews. And with the correct prompt, you’ll get meaningful content to showcase your impact and contributions to your organization.
As I mentioned in the introduction, I ended up using ChatGPT’s output as the backbone of my own review. It was an improvement over what I’d come up with originally: tighter overall and well organized. Because my first attempt at prompting here didn’t get great results, what I actually used was a combination of snippets from the final two prompts.
I thought there was a chance that ChatGPT would be more a partner than a replacement. I’d say it exceeded that bar. What I needed to add was some relevant context I’d not prompted the model with, some metrics, and a few other smaller fixes and tweaks. So while it’s not a full replacement for the entire process—you’ll have to collate your own accomplishments and smooth things out a bit—I believe it made my review better overall. It just took a minute or two to provide me with some really good snippets and a great draft to improve on.
Which is all to say: the next time I’m staring at a white screen fretting over how to start, I’ll head over to ChatGPT instead. If you have the same challenges I do writing your own reviews, I hope you’ll give it a try yourself.
Add a comment
Iterate on AI agents and models faster. Try Weights & Biases today.