Data driven design

Patrick Teunissen
Apr 4, 2018 · 7 min read

Back in the days
I can still remember the times where designing a website meant making the most cutting edge design for other designers. Even the B2B designs were high end design. When the design was finished, you’d just deliver the website to a developer with the simple message: “turn this into a website”. Did we check if the design was working? Did we check if they even liked the design? Did we check if there were places they got stuck or just plain left? Nope. But it looked good in my portfolio though. Designing desktop-first and wanting every page to get liked on Behance or by other designers.

Designers are arrogant
I used to be arrogant enough to let users follow my design instead of my design follow their needs and expectations. Good design tells you what you want and need, I thought. Silly me. They know as much as you do about what they want to buy, see or click. Of course we can steer a little in the direction we want them to go, but as soon as they notice they are not the one behind the wheel, they react.

For example, lately I see a lot of designers leaving out the scroll bar on desktop. I don’t even know why. But letting people get further than the first viewport is essential. This is where design causes harm instead of improving the UX and feel. Design should serve. It must be usable and shouldn’t force you to think.

The branding created for Kaartje2go isn’t the most cutting edge, from a designers’ standpoint. And for a good reason. The people who send cards are predominantly female, above 30 and are not necessarily looking for the latest designs, but rather wanting the personal touch and warm wishes. They just want to send a wish or paper hugs to a loved one. They are often not as experienced online as the generation after that.

They care more for a platform that works fluently, tells a positive story, and that we are honest and trustworthy. Far more important than to be fashionable. We let our products talk, our brand smile and our interactions love.

That sounds cute. But does it work? Prove it.

User Testing your design

We do user tests. These tests are usually for a specific category of cards, seeing categories have different emotions driving the users. In this case we tested pregnant women if and how they would find, create and order a card. They did this independently and were only asked something at the end of the experience. So as soon as they entered our site, they reacted to design, UX and colors.

Testing a color scheme
In previous user tests the mothers-to-be let us know our site didn’t look specialised in birth cards because we also had other categories of greeting cards. The design was too general. The brand colors, orange and blue, didn’t make it look as the site was for creating birth announcements.

We needed to look more specialised. Which we are. We quickly concluded we did not want to add more text of persuasion through USP’s or snack bars. The whole design needed to breathe birth.

Entering pages for birth announcements on will have a more suitable color palet instead of the orange & blue brand colors.

So in this user test we added a theme. The theme was a design overlay that gave the website more suitable colors for this specific category. Birth cards. People actually said they felt at the right place. The brand stayed recognisable but the experience was less distracting and the site felt calmer. Mission accomplished.

A/B Testing your design
Design is something that relies on emotion, usability and creativity. If something is creative it asks us a new question, shows us new perspectives, makes us feel something or makes us reflect on things. It can, sometimes, be very abstract and it often creates the same amount of people hating it as loving it. That’s when you know it is good design.

That sounds like a whole different game than data. You don’t want 50% to hate it, you don’t want people to get all philosophical whilst browsing or shopping and you certainly do not want to come across too profound or intellectual because that would exclude people.

Yet design and data attract each other because we have a lust for bringing them together. We want to know what we feel, why we feel it and leads us to understand ourselves better. That’s when sociology comes in. We can see peoples moves, what pages they visited and how long they stayed somewhere. This is science. Build your design on this.

So it might be a little weird designing on a data-based fashion. Yet, in my experience, there is no better way. A lot of my assumptions about people got crushed when data came back.

Data or it didn’t happen

A good example of this is an experiment we ran in the shopping basket. Our hypothesis was that too many options in the shopping basket breaks the users concentration and focus on the task at hand; check out the order. So our solution was creating a less complex basket page where we brought it back to its core. Not everything was independently editable anymore. Summarise and check out. Fewer options, more clarity.

Control: version where all prior chosen fields are still editable via the pencil icon
Variation: the basket is made cleaner by removing all options separately editable and replacing it with just 2 buttons.

Goals: Creating less confusion and see more checkouts.
Possible downside: the specifications were no longer editable so that’ll increase questions for our customer happiness team.

Results: The orders never increased. Yet the average order value grew significantly. We realized people stuck to their original choices. As they didn’t see the overall total, it could be a little above budget. Giving them the option to go with a cheaper order apparently triggered them to do so. As soon as you delete the options they actually order what they wanted. The details of the order could still be changed, but not in a single click. And more importantly they didn’t necessarily expect the options there so it didn’t result in a less positive experience.

Although the ‘less is more’ tactic worked, it told us something more valuable; people already made a choice on what they wanted to order and letting them ‘choose’ again created doubt.

Clear your browser history

A lot of things we designers do is instinctive. I noticed this when interns started asking questions I no longer asked myself. Why does this font work? Why did I increase the padding? I had to reverse engineer all the reasons of the design choices I made. I automated a lot of design work. The interns, on their turn, showed me what I did on auto-pilot and where did the rebel designer came in to play. A very good lesson.

A lot of my decisions were based on print, old and new design rules. These rules were commonly expected in web design, yet if it really worked for our users was unclear. And I do not think I have to tell you that rules are meant to be broken, right.

Highlighted are just a glimpse of the elements tested on the category page; filtering box, category image and the 3 USP’s within the fold.

I like to see pages that are stripped down to their very purpose. So if our head marketing wants to experiment by adding yet another USP somewhere, I immediately think it won’t result in more of anything. But it can. Design is not leading for our users. So reading a whole bunch of unique selling points let’s them make a decision. This tells me the users respond differently than the average online shopper to web & UX design.

The moral to the story: Never assume anything, especially as a designer. Assumption is the mother of all fuck-ups. Really try to investigate what your target group does or doesn’t like. Do not let your years of experience make you into a dictator. More importantly: don’t let it forget your primary goal, which is to create a very usable and effortless shopping experience. So if it feels good or looks nice can’t be the only the reason to do it. I know it too well, that gut feeling of designing something good. But a gut feeling is testable. Validate your designs through tests and experiments. Also a lot of researches are already been done on the topic, so get the knowledge on the world wide web, too. This will give you an insight in your demographics’ wants and needs.

Thanks for reading this and hopefully you could take away something from this!


Tech @ Kaartje2go