Job Search Sankey

Background & Purpose

In October 2024, I began applying to jobs intensely after learning about an impending layoff. Like many job seekers, I found myself sending out dozens of applications, often with little visibility into how far I was getting in the process. By early 2025, I had resumed applying again with a renewed focus and a more structured approach to tracking progress.

To bring clarity to this overwhelming experience, I began logging each application I submitted, along with the specific stages it went through — from initial submission to final outcome. I wanted to know:

• How many applications progressed beyond the first step?
• At what stage were most rejections happening?
• How many companies made it to final interviews or offers?

This notebook uses Sankey diagrams to visualize the flow of applications across various stages in the interview process. Each node represents a distinct stage (e.g., "Applied", "HR Call", "Technical Interview", "Offer"), and the links represent transitions between them. The thickness of each link corresponds to the number of applications that followed that particular path.

By visualizing this data:

• I can see common patterns and dead ends.
• I get insight into which stages were the most common points of rejection.
• I gain a clear picture of where effort translated into progress — and where it didn’t.

This project is both personal and analytical: a way to make sense of a challenging time and derive insights that could help improve future job searches.

Example output

This Sankey diagram visualizes the flow of job applications through various interview stages. Each node represents a stage in the process (e.g., Applied, Recruiter Inquiry, Technical Interview), and the width of the connecting arrows indicates how many applications moved between those stages. This visualization helps identify common paths, bottlenecks, and outcomes in a job search.

What it Does

• Visualizes application flow through various interview stages (e.g., "Applied", "HR Call", "Technical Interview", "Offer").
• Quantifies and displays how many applications moved between each stage.
• Helps you identify drop-off points and common paths through the job search process.
• Tracks transitions by company, stage, and time period.

Getting Started

1. Clone the repo
2. Set up a venv or Python Kernel in Jupyter Notebooks
3. Install dependencies in the requirements.txt file:

pip install -r requirements.txt

4. Update user input values by creating a .env file
5. Run all cells in the notebook

Data

The input is a .csv file containing transitions between stages for each job application. Each row represents a source → target transition (e.g., "Applied" → "HR Call"), along with company name.

The input file format is expecting the following columns:

Company	Source	Target
Company1	Applied	HR/ Hiring Manager Interview
Company1	HR/ Hiring Manager Interview	Technical Interview
Company1	Technical Interview	Panel Interview
Company1	Panel Interview	Offer
Company1	Offer	Offer Accepted

SOURCE_COLUMN: The initial stage of each application step (e.g., "Applied")

TARGET_COLUMN: The next stage (e.g., "Technical Interview", "Rejected")

Users can upload a CSV file with any column structure, but must specify in the configuration which columns represent the source stage and target stage to generate the Sankey diagram correctly. Other columns can be included in the file, but will be ignored in the analysis.

Configuration

To customize the visualization for your own job search data, you'll need to configure a few values. This project uses a .env file to store those settings.

Step 1: Create a .env file

Start by copying the example configuration:

cp .env.example .env

Step 2: Configure environment variables

Then open the .env file and fill in the required fields based on your CSV file:
INPUT_FILE=your-data.csv # Path to your CSV file
SOURCE_COLUMN=Source # Column name representing the starting stage of a transition
TARGET_COLUMN=Target # Column name representing the next stage of a transition
VIS_TITLE=Job Search Sankey Diagram # Title to display on the Sankey diagram

Technologies

• Python
• Pandas
• Plotly (for Sankey diagram generation)
• Jupyter Notebook

Stage Transitions

The program can handle any interview stage transitions you provide. However, to analyze and compare progress effectively, it’s important to use a consistent set of stages and keep them in the same logical order, even if the actual steps happened in a different sequence. This consistency allows meaningful comparisons across different companies and application processes. Below are the specific stages I used in my analysis, based on my personal job search experience:

Possible Mappings

Applied/ Recruited

• Recruiter Inquiry → HR/ Hiring Manager Interview
• Applied → Auto-Rejected
• Applied → HR/ Hiring Manager Interview
• Applied → No Response
• Applied → Withdrew

Hiring Manager Interview

• HR/ Hiring Manager Interview → No Response
• HR/ Hiring Manager Interview → Rejected
• HR/ Hiring Manager Interview → Offer
• HR/ Hiring Manager Interview → Technical Interview
• HR/ Hiring Manager Interview → Withdrew

Technical Interview

• Technical Interview → Panel Interview
• Technical Interview → Rejected
• Technical Interview → Withdrew

Panel Interview

• Panel Interview → Withdrew
• Panel Interview → Rejected
• Panel Interview → Offer

Offer

• Offer → Offer Accepted
• Offer → Offer Declined

Use Cases

• Identify common failure points in the application process
• Compare how companies handle recruiting
• Visualize application volume by stage or over time