So when people are presented with something which is visually appealing, we think it's easy to use, even when it isn't. And people will then default to blaming themselves, not the pretty, elegant thing, because clearly the pretty elegant thing isn't the issue.

We call this the aesthetic-usability effect. Perception of the expected experience, and attribution of the actual experience, is more important part than the actual experience.

It's one of the many ways in which engineers, economists and analysts (in my experience) tend to run in to issues. They want people to behave rationally, based on their KPIs, not as people actually experience and interact with the world.

There's all sorts of research that then comes off this, like people enjoying wine they've been told is more expensive, over wine they've been told is cheaper, and the physiological response as measured with an MRI confirms their reported divergence in experience, despite that the wines are the same, as one quick example.

Low contextuality evaluations (my term for where you ask someone to state things about something where they lack enough experience with enough breadth and depth to answer reliably) are always wonky. People can't comment on wine, because they don't know enough about wine, so they seek other clues to tell them about what they're experiencing. Similarly, people don't know about things that are new to them (by default) or that look different to what they expect, so their experience is always reported as being worse than it probably actually is, because their brain doesn't like expending energy learning about something new. They'd rather something they understood. It's where contextualisation and mimicry come in really useful from a design of experience standpoint.

Before we had that latter data I actually argued against attempting it - I figured having a clear usability win vs iPhone would be an area we could capitalize on, and didn't believe we'd be able to execute the swipe system well in the time we had (I'd rather be behind and robust than leading edge and flaky), but doing it was definitely the right call - felt pretty sheepish about that one for a few years. The eng and ux teams that pulled it off were next level.