Lesson learned looking at more than one metric to inform your designs - Stephanie Lawrence
UX designer Stephanie Lawrence explains how to look at relevant metrics to inform UX design work—and how optimizing page elements in isolation, without understanding the larger picture, can be counterproductive and actually result in fewer sales.
What Stephanie covers:
- Why it's important to understand the larger context of a customer's actions
- A practical example of when optimizing in isolation could have been counterproductive
Click below to read the transcript.
Lesson learned looking at more than one metric to inform your designs - transcript
Hi, thank you so much for watching this talk. My name is Stephanie, and I'd like to share a lesson that I learned while working as a UX designer on the eCommerce site called Uncommon Goods.
So I was working on a gift finder, in particular. So you go through, enter some information about who you'd like to get a gift for, and then you get some suggested results, and can go from there, favorite items that you like, look at reviews for it, things like that.
And one thing I realized is that, yes, I could look at all of the data being collected, all those particular metrics, like individually and see, okay, which ones are important and optimize for them accordingly, and there'd already been some really important KPIs identified.
For example, around people favoriting items, the more items they favorited, the better, in terms of them purchasing something and the same thing for them actually looking at the reviews for the items that they would get in their results, right? So you think, all right, let me go through and optimize for that and, you know, basically aim for seeing those numbers go up.
But what I'd like to argue for is that there's another approach where you look at these metrics in relationship to one another.
And in doing so, you're trying to better understand the context of the actions that they're representing within a customer's experience.
And in turn, use that to kind of understand the story that that data is telling about the different experiences that customers are having, and then see which one of those experiences is a good indicator of purchase and not just one metric alone.
So I'd like to show you an example of that. And this is based on a real world example. And I've just simplified it for the sake of time. But in this case, again, like I said, I was working on this gift finder. And again, looking at how many items the person favorited in the session and how many reviews they looked at. And I thought, I want to look at more metrics together because, say for example, if someone gets only five results from the gift finder, them favoriting an item is a lot different than if they got 50 and they decided to only favorite one, right?
So I went through and worked with the analytics team. And I thought to myself, I won't just want these numbers to go up, what do they actually mean in relationship to one another? For example, the amount of results that someone sees versus what they're favoriting.
And I did look at those results, and again, this is a simplification of it, showing the averages of each one of them, and worked with the analytics team to basically group the customers that we collect the data on based on similar patterns of behavior. But at the same time, you don't have to do that. You could take a much more simpler approach if you have different customer segments that you're already looking at and go from there.
But point being, I took a much more statistical approach here. And then I looked at the relationship between these metrics within each of these groups, and group three is highlighted because they're the group that bought something. So, there are three groups being represented here that had similar behaviors when they were going through the gift finder. And the third group's the only group that bought something, but they are not the ones that favorited the most out of the three groups, and they're not the ones that read the most reviews out of those three groups.
And those are important KPIs that indicated, hey, so this person will probably buy something. But there's one significant thing, they saw less results than the other groups, right? So in turn, then you start thinking about it a little bit differently. It's like, hmm maybe, like, what story is being told? Someone's getting some results, and they like a good amount of the ones they see and they read a good amount of those results, and then they go on to purchase something.
As opposed to maybe the other two groups where they saw a ton of things, didn't like as many of them out of the results that they got, and both represented in their favoriting and those reviews, and in turn didn't buy something. So, that's the story.
So, if I'm thinking about the experience of someone who actually goes through and purchases something, I'm thinking about someone who likes a lot of the results they get and also reads reviews for them, and thinking about the percentage of items that they favorite out of the results that they're getting, rather than just looking at them, favoriting, just looking at that average number of favorites going up and the average number of reviews being read. So that's what I'm talking about.
And, in doing that, you start to map out stories, like I said, that person who's going through and liking a lot of the results they see versus someone who goes through and doesn't like as many things, and that's based on data. But that's a qualitative analysis that can then inform what decisions you might want to make later in the design process and what kind of experiences you want to optimize for. So I hope that this was useful. I hope that it was helpful, and I hope that you try it out in your design process and your design practice. And thank you so much, and please feel free to email me or contact me on Twitter if you have any questions.