Why Do Startups Fail? A Closer Look at the Semantics

Luca Scagliarini - 14 September 2015

Luca Scagliarini, CEO, Expert System USA

– When nine out of every ten startups fail, how can a creative team of ambitious entrepreneurs improve their chances to succeed? Autopsy.io is a site dedicated to failed startups and aims to help new startups avoid the mistakes made by others before them.

In an effort to gain insight on key themes around failure, we used our expertise in cognitive computing and Cogito context analysis technology to analyze the data submitted to Autopsy.io. Our Cogito technology reads, understands and analyzes text as humans do, so we are able to provide intelligent insights into the challenges new startups encounter.

Leveraging the power of semantic technology, we are able to automatically generate a fast and accurate understanding of text in its proper context and make sense of the information at hand. With Autopsy.io we determined the top 10 concepts cited when describing failure were Company, Product, Investor, Startup, Money, App, Team, Website, Market, and Client. Surprisingly we found that some common business factors such as marketing, revenue and employee issues were not among the top 10. Further analysis of the specific words in sentences including the words “fail” or “failure” revealed product (15) was mentioned far more often than employee (2), marketing (2), sell (2), and revenue (2) signifying that these factors were not mentioned as predominant reasons for failure.

When we looked further into some of the most common reasons for failure, we found a correlation of how often these key topics were brought up or not even brought up at all in the post mortem discussions around startup failures. The below data shows that more entries on Autopsy.io include the words money, capital, fund, investor, product, team, and idea, when describing reasons for failure compared to not mentioning these terms at all.

Mentioned Not Mentioned
Money or Capital or Fund 67 19
Fail Or Failure 66 20
Product 66 20
Idea 65 21
Team 55 31
Investor or VC 50 36


When analyzing the post-mortem comments to determine the specific reasons for the eventual demise of these companies, entrepreneurs were quick to point out that it was a particular problem (196) that led them to eventually fail (137), failure (95) or failed (28). Additionally, individuals often recognized their failures admitting to particular mistakes (92), when the business became too difficult (59), or they made a wrong (39) decision.

When examining the descriptions these entrepreneurs used to indicate what they were trying to do, four of the top five action words used had to do with creating an actual product (make, build, work, want, start). These failed startups sought to create something new, but did they execute? Words such as launch (102) and sell (80) were mentioned more frequently than deliver (27) and execute (18) suggesting perhaps the aspiration was much bigger than the actual execution.  Interesting enough, comments that focused on implement (12), integrate (11) and enable (12) ranked toward the bottom, possibly indicating that many were far from realizing their goal.

If you build it, will they really come? Many failed startups have proven that the ‘field of dreams’ concept rarely works when it comes to taking a product to market. While product (454) was mentioned quite often in the commentary, customers (202) and market (176) were mentioned far less. You can potentially conclude that there may have been less focus on a true market demand and customer needs and more focus on the internal development of a product.

While the Autopsy.io site is still fairly new, these collective insights can serve as a valuable learning source for entrepreneurs as they set out on their startup journey themselves. We will continue to monitor Autopsy.io as we believe its value will increase as more startups are willing to share their stories and lessons learned.

To read the entire Autopsy.io report, please click here.