The Feedback Coach

• 

  The Feedback Coach Logo

• 

  1. Write feedback in your own words and input it into the app

• 

  2. Once you've submitted your feedback you'll receive an analysis, as well as guidance on how you can improve

• 

  3. Once you've read the analysis and guidance, try putting the tips into practice by editing your feedback

• 

  4. Need further guidance? Take a look at the information on effective feedback theory

• 

  5. Happy with your edited feedback? Ask The Feedback Coach to review in the 'chat' box. See where you've improved and what you've learnt!

Inspiration

Feedback is a gift, but it can be difficult to learn how to give feedback that will be well received by the recipient. Likewise, it can be difficult to understand the true intent of feedback if it's not delivered in the right way. Have you ever asked a friend or colleague to review written feedback before you send it to the intended recipient? This app acts as a trusted advisor to do the same thing.

What it does

The Feedback Coach helps you to craft effective, professional feedback for colleagues. It will review feedback you have written and identify any potential issues such as bias or language that could elicit defensiveness or offend the recipient. Rather than just providing an improved, edited version of the feedback, The Feedback Coach will build your skills by explaining best practices and theory behind giving effective, professional feedback. This helps you to gain confidence in situations where you need to deliver in person feedback, where you might not have access to the app for support.

How we built it

The Feedback Coach started as an internal Amazon tool, which I built to support Amazon's internal performance evaluation process. As part of the process we provide feedback to colleagues on their superpowers and growth areas. As I was writing the feedback I was conscious of the impact my unconscious bias may have and wanted a way to identify this before sending the feedback to colleagues. I decided to experiment with PartyRock and created an app to review feedback based on Amazon's principles. This app worked really well and eventually I demonstrated it to my colleagues. My colleagues' feedback made me realise that there is a use case for a similar app covering a broader range of feedback scenarios. I revisited PartyRock and used the 'App Builder' prompt to describe my ideal feedback coaching application, which PartyRock quickly built for me. Once I had the initial app I was able to adjust some of the prompts, add references to other widgets and experiment with different LLMs to tailor the app to my requirements.

Challenges we ran into

My role is business (sales) focused and so I don't have a deep technical background. PartyRock enabled me to quickly make an idea reality. As part of building PartyRock I've needed to familiarise myself with GenAI concepts including LLMs. Due to time constraints I haven't experimented with all of the PartyRock features, such as the 'temperature' feature. Increasing the temperature or Top P increases the randomness of the model's response, resulting in more creative and imaginative text. At this stage I have kept temperature low so that the response is more factual, but with more time I may be able to experiment with this further. One idea I considered is that the app could automatically edit the feedback which would save the user time. This is something I included in the previous feedback app I built. As the goal for this app is to provide coaching to help build the user's ability to provide effective feedback I have decided not to include this feature at this stage. Manual editing encourages the user to review the analysis and feedback theory in more detail, which helps to improve the user's knowledge around how to deliver effective feedback.

Accomplishments that we're proud of

Building this app has made me realise the potential for low code / no code development. I'm proud that I have been able to make my initial idea a reality using PartyRock and I'm excited for the potential this has to improve our ability to give and receive valuable feedback.

What we learned

I experimented with different models and found that Claude is most effective for the desired use case. Jurassic-2 provides a more concise summary of the feedback, which would fit other use cases well but in this case doesn't give enough detail to act as an effective coach. So far I've only worked on this alone but I would be open to suggestions from others to help improve the application further.

What's next for The Feedback Coach

Research has shown that creating a feedback culture is the most critical driver of positive organizational and financial outcomes, however only a third of employees report actually receiving the feedback they need to engage and improve. Feedback can be uncomfortable for many people, regardless of whether you are on the giving or receiving end. In fact, one study found that receivers’ heart rates jumped enough to show moderate or extreme duress in spontaneous feedback scenarios—and both the receiver and giver of feedback felt equally anxious.

I'd like to see The Feedback Coach being adopted by people managers to improve how they deliver feedback to their team in a thoughtful way and as a tool to help their teams learn how to deliver feedback effectively. I also think it would be great to see The Feedback Coach built into performance evaluation tools and processes so that people can check their written peer to peer feedback before submitting. On a personal note I'd like to become more confident with giving constructive feedback and this app will help me to achieve that. By leveraging The Feedback Coach we can create a happier, higher performing workforce and facilitate a culture of respect.

Built With

• claude
• partyrock