It’s now possible to conduct user testing without humans.

A startup called Synthetic Users says it has reached 85% to 92% parity scores with real humans.

To do calculate this score, it takes the transcripts of user research interviews conducted with real humans and compares them with the transcripts generated from interviews with synthetic users.

It takes into consideration:

• Overlapping themes

• The depth and specificity of insights

• Comprehensiveness of coverage

• and Qualitative alignment

A comparison score is then created to understand the differences between the two.

It’s still early days, but the time spent conducting user research could drop drastically in the future.

I could see this being very helpful for augmenting traditional research - but (hopefully) not replacing it.

In the latest Knowledge Series, I explore more use cases for synthetic data: departmentofproduct.sub…