Data Analysis on Factors that Affect Success of a Startup
The skills I demoed here can be learned through taking Data Science with Machine Learning bootcamp with NYC Data Science Academy.
Introduction
We have seen startups, such as AirBnb, Uber, Twitter, etc. scale into technological giants. At the same time, there have been so many that haven't been able to create their niche in this competitive industry. Keeping up with the trends and staying ahead of the game is very important for someone starting a business. At a time when the smallest of ideas have the potential to disrupt entire industries and niches, it is crucial to be prepared. Many startups leave their mark because of their impressive business models and strategies. Based on the data, I will be studying the factors that seem to have a strong correlation with a startup's success and analyzing these trends.
Objective
By understanding the factors that prove to lead to a startup's profit, a founder can better develop their business practices, team structure, investment strategies, etc. I decided to find out what these factors are and providing more insight into what these dependencies look like. For this, I obtained two datasets from Kaggle.com that consisted of information about 200+ startups, their status, and their reasons for failure / success. I decided to study this data from 4 different directions which are:
- Industry Analysis - What are the more successful industries and is there more likelihood of startups in a certain realm to succeed?
- Team Structure & Founder's Background - What does the background of the founders and employees look like? Do larger startups have a bigger shot at being profitable? What are the demographics?
- Product / Service Offering - What kind of product / service is the company offering? Are certain technologies more popular?
- Investments and External Factors - How do finances and investments factor into the success of the startup?
To identify these dimensions of analysis, I studied the reasons of failure that these companies provided. The following word cloud allowed me to understand what the common problem was amongst these startups.
From the Wordcloud, I was able to better anticipate what the key problems may have been. I further analyzed these reasons to understand what the distribution was. This analysis solidified my hypothesis that product, market, funding, business model, and lack of focus were the root causes of most failed companies. Let's further examine the data from each of these facets.
Industry Data Analysis
To study the success of companies across industries, I first tried to dissect how many companies of different industries have been launched across the years.

From this, we can conclude that there is a significant growth in the number of startups launched every year. It is also evident that a huge percentage of the initiatives were Marketing & Sales focused products or services. This gives us a better view of the crowding in each industry.
I hypothesized that those industries with many emerging startups would be more likely to fail due to high competition. To corroborate this theory, I looked into the trend of success / failure in each one of these industries across the years. The following graph shows the trend of success of Marketing / Sales companies across the years and it seems that the success rate increased over the years. So, industry doesn't seem to be a large factor in the success of the initiative.

Product / Service Data Analysis
Now, I believe that the product or service that a startup is offering is one of the most important factors of success. The evolution of technologies over the years could also be a contributing factor to the success of startups. But, the following charts show that Machine Learning & Cloud based businesses were less likely to fail than those that were not using machine learning or cloud technologies. One can also conclude whether a startup is product or service based, it doesn't affect its chance of success.

The following chart also shows what the distribution of the failed and successful companies were based on the Gartner Hype Cycle Stage that they were in.

From this, it seems that those companies that were in the peak stage of the Gartner Cycle had a better chance of being successful. This would imply that the technologies that they are using are a bit more developed and that they are very popular and sought after in the market.

From this analysis, we can conclude that the number of years of experience that one has increases the chance of their business being successful. However, their level of education or whether they have been a part of successful startups in the past doesn't correlate in any way.
Then, by looking at the number of employees that a startup has, I am able to make some conclusions. The distribution for success and failed companies looks the same regardless of the number of employees, so this indicates little correlation. I also looked at the correlation of success with the team composition score of the startup. This score represents the diversity of the team. The percentage of companies that fail with a high or medium score is very low.

Finance & Investments
Lastly, finances play a big part in the establishment of a business. It is very difficult to execute a vision without some monetary investments and finances are a big reason for these companies to fail. When I look at the distribution of the amount of funding that these companies received, it is evident that the funding amount and chance of success have a strong positive correlation.
In conclusion, I was able to identify some very interesting relationships between various factors and likelihood of success in the startup scene. I think that in the future, expanding this study with a larger set of data could prove to really provide some interesting insights.
To use the interactive app and look at the data yourself click here:
https://sharanyaagarwal.shinyapps.io/ShinyApp/
Github repository:
https://github.com/sharanyaagarwal/DataScience/tree/master/ShinyApp