I was always interested in exploring the trajectory of startups across the globe. I first got interested in this because of my own startup venture that I launched with a couple of my friends in India. Being on a founding product team, I remember making decisions with limited information and constantly worrying about how the results would turn out. I started speculating about what it takes for startups to take off and survive. I finally got to explore a part of this question through this Information Visualization project. 

I came across a dataset that collected information from over 66,000 companies founded between 1996 and 2016 across 100+ countries. The dataset found on Kaggle, originally sourced from Crunchbase (one of the most widely used startup databases in the world), was packed with information about funding, founding dates, geography, industry, and outcome status, i.e., operating, acquired, IPO. 

Note: While working with this data, I noticed some records were incomplete, mostly related to founding dates and funding information. For simplicity, I chose to remove incomplete data from the dataset while building my visualizations.

Visualization 1: The startup boom – How the global startup wave grew from 1996 to 2016

First thing I wanted to understand was “What led to startups being so popular?” and “When did the startup ecosystem actually take off?”

While the dot-com peak in 2000 and the dot-com crash in 2001 were significant events, I was surprised to see that the actual growth of the startup economy happened much later in 2004. The curve keeps growing over the next 10 years to reach an all-time high of 6,424 startups in 2012. I couldn’t figure out the reason for the increase in startups over these 10 years or the crash that can be seen immediately after. I suspect the decline in numbers after 2012 is potentially due to a lack of data. 

Initially, this graph was kept simple, but after conducting user research, I realized that viewers wanted to know more about the events that led to the incline or decline in numbers over the years/ Hence, I added annotations to the graph to make it more relevant to my target audience. 

View Visualization 1 in Tableau

Visualization 2: Global Startup Density. 

After understanding the timeline of the startup economy, I wanted to understand where most of the startups came from. I created a map to visualize this information, and the result was pretty eye-opening. The USA came up on top with the largest number of startups (37,601), followed by the UK (3,688), which is an impressive 10x gap! 

While designing this map, figuring out the color scale took me a while because the number of startups in the US was too high. This led to all other countries’ counts being regarded at the same level. The map ended up looking monochromatic, with the USA as the only country with the darkest shade. I addressed this issue by using stepped color bands and capping the color scale at 2500. 

View Visualization 2 in Tableau

Visualization 3: Where do startups go public?

The third important question looming in my mind was about outcomes. What happens to these startups? Do they still exist, or have they collapsed? I learned that not all startups may have the same kind of exit. Some startups continue to operate, some get bought by another company (acquired), some go public (IPO), and others shut down (closed). The number of ‘operating’ startups was irrelevant to my visualization, because I noticed that most of the startups were in the ‘operating’ stage, and it took up 80% of the visualization’s real estate. So I decided to remove that category completely and display startups that are in the ‘acquired’, ‘ipo’, and ‘closed’ stages. 

This interactive chart shows the fate of startups across the top 10 startup countries. Viewers can sort countries by outcome to see which country is leading in startups acquired, public, or closed. This visualization can provide viewers with important information about the success rates of startups in these countries. 

The sorting mechanism itself was a little difficult for me to figure out. In earlier versions, while sorting by count rather than rate, the USA always turned up in the top because it was leading in the number of startups. To fix the issue, I created a calculated field that responds to the parameter dynamically. 

View the Interactive Version of Visualization 2 in Tableau (recommended)

Reflection:

After creating the visualization, I realized that even though my data was taken from Crunchbase, there were significant flaws and missing information in it. This led me to question its authenticity and ask myself, “How do I validate if my dataset is accurate?” 

Another limitation of the dataset was that it had information for startups between 1996 and 2016. Next time, I would want to work with more recent data to see how the pandemic affected the startup ecosystem.

Personally, I feel there are still a lot of questions that I haven’t been able to answer here. More specifically, I wanted to uncover what makes a startup ‘successful’. In the future, I would explore the funding information in this dataset to understand the role it plays in building successful startups. 

Exploring Tableau without any prior experience was certainly challenging. I learnt a lot of new operations on the tool, but initially I struggled to figure out how to get the result I desired. I want to acknowledge here that I used AI (Claude) as a tutor to help me navigate certain complex functions. Troubleshooting using AI to understand how to use Tableau properly taught me a lot along the way.