A Product Manager Explains the Majority Illusion

During last year's presidential election, you may have noticed that everyone on Facebook either:

  • Hated him or...
  • Hated her

And thus you assumed that your favored candidate must be headed for victory!

Because everyone thinks so... right? (Everyone = everyone in your Facebook network/ Twitter network/ Instagram network...)

majority Illusion 2.jpg

I also suspected I knew exactly who would win the presidency based on what I saw posted in my social networks. And yet, I believe in the power of data.

So I ran some big data queries during the election. What did I learn?

  • Trump was mentioned on social media 3-4x more than Hillary (depending on the day)
  • When run through a sentiment model, the percentage of Trump's negative social media mentions were only was only *slightly* greater than Hillary's
  • What does this mean? - Trump still had millions (and millions) more positive social media mentions than Hillary overall. 

I don't want to get into politics, but this blew my mind. I could not imagine where all these Trump voters lived. "Where the hell are they?" I wondered out loud (on an almost daily basis as I watched my queries continue to return data).

And yet I know that data doesn't lie. So I was pretty confident in who was going to win the election.

This is a great example of the Majority Illusion.

The majority illusion is: the phenomenon in which an individual can observe a behavior or attribute in most of his or her friends, even though it is rare in the network as a whole.

How does The Majority Illusion apply to product management?

Statements I hear often as a product manager: 

  • "Everyone thinks X feature is top priority."

  • "Everyone thinks the product should work this way."

  • "Everyone thinks X Feature is more important than Y Feature."

When discussing new features or products, I've often heard engineers and product managers alike exclaim, quite confidently, that everyone will use Y feature this way.  Or that everyone wants X feature before Y feature. This may not be true! Without user data, the person making this statement is probably biased by the people in their network, and their (limited, non-randomized) feedback on the product or desired features.  

When prioritizing features or considering a UX change, it's important to take a data driven approach and avoid falling prey to the Majority Illusion. Talk to customers, and also look at usage analytics for your product. Talk to sales people, and engineers. Do not be blinded by what your immediate social network is certain should be done.

Further Resources: