A time when i designed a dashboard, no body returned to.

In 2023, i designed a dashboard which no one returned to. Some users even said better don't give me a dashboard, it will add more to my daily work.

My users were school counselors and they were already overwhelmed by their workload. The dashboard i am talking about was the one to show different data based on the student responses which our student app collected during online lessons.

Palo is a SEL (Social Emotional Learning) platform focused on middle schoolers of USA. Students used our app to complete one bite-sized interactive lesson through quizzes, video lessons, puzzles and by expressing their feelings on several topics. There were mood data and lots of text responses.

So counselors need to see what is the mood of the school and find students who needed their intervention. I applied all my knowledge and failed to make a dashboard that worked for them.

Then i decided to stop iterating the dashboard and take a 'Fail Fast' approach to understand the true intent. I decided to join them in their workflow first; that is their mailbox.

Used Canva to make weekly reports and see the engagement. Iterated fast without the dev loop and eventually realized the real need.

We needed to flip the dashboard flow. Instead of data first we needed action first. What counselors needed was a list of student that they should focus today, and the reason why.

Thats were we decided to use AI (OpenAI has just release API) for the first time and analyzed all responses to create a list of students which needed attention.

We created a dashboard which had just names of the students who needed attention today and nothing else. Counselors can click any name and see trends about the student and why the student is on the list.

So our dashboard's KPI was not 'how much time user spent here' but 'how less they can'. Palo was used in 12+ school districts then and counselors reported ~60% reduction in their review time and they kept returning.

This exercise gave me the first hint AI will change how we can interact with data and what people will be doing mainly in AI native environments. I am curious and constantly experimenting and shipping this new experiences.

So main lesson- Fail fast, iterate faster with or without vibe coding; you got to find the right way to reach the truth. Its not always the fast shipping but understanding a users typical day and getting in their shoes is what matters to truly ship faster and deliver true value.

Here is the full case study.