Project — NPS Analysis

upraised.co

Applied Python & Google CoLab

to analyse 300+ feedback, developing hypotheses for improving NPS.

Applied Python & Google CoLab to analyse 300+ feedback, developing hypotheses for improving NPS.

Project Duration

Sep - Nov 2023

Role

Product Designer

Collaborators

Upraised, India's upskilling Ed-tech platform for product management roles, operates in a cohort format—guiding users through Learning, Interviews, and Job placement.


In this project, I worked on analysing the NPS data collected from period of Jan to Jun 2023 to find the areas to improve product.


(For every user, we share a NPS form to collect feedback on different aspects of program)

Why work on improving NPS?


In the learning and placement industry, trust relies heavily on word of mouth and user experiences, making the Net Promoter Score (NPS) a crucial metric.

Aug 2023

Aug 2023

Stage I — Cleaning & Understanding data

The NPS data was highly textual and have to start with defining themes, converting the textual data into themes.

For Example


The Question: How are you feeling?

Possible Themes: Upraised, Non-Upraised, Positive, Negative, etc.

Stage II — Plotting Graphs

Used Python to plot various graphs -

Plot 1 - Theme vs Avg. NPS score

This helped to identify the impact of each theme and gave clear visibility on which theme can be probed for further analysis

Observations

  1. U_Negative has given NPS avg of 7.5

  2. NU_Neutral and U_Positive have NPS avg of 8.5

  3. NU_Positive and NU_Negative don’t have much variance in terms of the NPS avg.

Plot 2 - Distribution of themes by NPS Segment

This helped to identify the percentages of promoters, detractors and neutrals in the themes. I can target areas where there are abnormally large percentage of detractors.

Observations

  1. Over 50 percent of detractors come from U_Negative

  2. Over 40 percent users in O_Neutral are detractors

Plot 3 - Impact of the themes

The difference between the overall NPS and the NPS excluding the selected theme is the impact of that theme.

Observations

  1. U_Positive immensely positively impacts the NPS score.

  2. U_Negative negative impacts the NPS score.

  3. NU_Negative, NU_Positive and NU_Neutral have loose co-relation with the NPS across.

End of Stage II

End of Stage II

We observed that we should look into textual feedback tagged with U_Negative and O_Neutral to find further insights.

Stage III — Themes and formulating Hypothesis

What’s the recurring theme in the U_Negative?

  1. Overwhelmed by the program (and work) - Count 16

  2. Not feeling confident about product skills (learnings) - count 18

  3. Not sure about getting a job at the end of course - count 8

  4. Support response should be faster - count 3

The main theme is “User is overwhelmed...” due to the work and program and they feel anxious about solving all the pending work in the given deadlines.

Writing Hypotheses

I repeated the Stages I to III for 5 different questions and then made a master list of hypotheses to help improve the NPS score.

Next Case Study

How I Improved interviews UX for 1000+ candidates, totalling 5000+ interviews

Led design of program dashboard, reducing tracking time by 50 hrs/week

Led design of program dashboard, reducing

tracking time by

50 hrs/week

Project timeline

Mar 2022 - Nov 2023

Mar 2022 - Nov 2023

Reading time

6 mins

Company

Upraised

Upraised