AI is not yet good enough for competitor analysis

One of the most basic tasks in competitor analysis is finding the competitor’s customer list. There are websites dedicated to aggregating such lists, and most software companies tell you at least some of their customers on their website.

How does AI do at this task? Let’s take a random software company, Coveo - which lists dozens of customers on its own website:

We will ask the AI to look on coveo.com and also on websites that a human analyst would check when building a customer list:

  • https://idatalabs.com/

  • https://trendspro.builtwith.com/

  • https://nerdydata.com/

  • https://publicwww.com/

  • https://enlyft.com/

  • https://www.similarweb.com/

  • https://www.google.com/search?q=coveo+%22case+study%22

  • https://www.cloudwards.net/

  • https://www.appsruntheworld.com/

None of the AIs did a good job with this.

ChatGPT says it cannot do the task:

But when asked, ChatGPT recommends Copilot or Gemini as able to do the job - let’s check.

Copilot says it cannot do the task (its claims about sensitive information are nonsense - Coveo has a public list of customers) even though it seems to know that at least one of the sources (coveo.com itself) does have the information:

Gemini finds some customers on Enlyft, but claims there is no other source even though coveo.com lists dozens of customers:

This is disappointing from major AI platforms. An intern would have done a better job - because an intern would have found the dozens of examples on coveo.com. There is a long way to go for AI.

 

Need a quote for competitor research?

 
Previous
Previous

Make the business case for competitive intelligence

Next
Next

What is competitor based pricing?