Font Feedback Experiment

I often get feedback on my articles, and try to respond either directly or via follow up posts. This is one of the reasons I have so many articles on pie charts (you only have yourselves to blame!) A few weeks ago I received an anonymous bit of feedback about the typewriter font I use on my website.

Hi there. I enjoy the content of your website but find that the typeface makes it very hard to read. It would be really nice if there was a way to turn off the distressed web font on blog posts. Sorry for the unsolicited website critique. Just figured I’d let you know.

I have worked in many different places, and have witnessed the alarming defensive responses to the kind of feedback that you may have encountered yourself from time to time. Things like “they don’t understand the design”, or “don’t they use a browser with reading mode”, or “I’m in charge, so I’m not changing anything”.

I am not this kind of organisation; I like feedback! I like feedback even more when it means I can run an experiment.


I could run a usability test and ask people how they felt when they viewed different versions of my website, but people in usability tests aren’t really trying to find out why you shouldn’t mix TypeScript namespaces and modules, or why if-statements provide clues for better code design. The people who are trying to achieve such things are real users. The other problem with usability tests is scale. I could probably convince 15 people to provide feedback, which is a small sample.

So I turned to Google Optimize. Not only does it allow me to get feedback from people visiting my website with real goals in mind, it also gives me way more than 15 people.

So I set up three versions of my website, using three different fonts:

  • The current font
  • A slightly less grungey version of the current font (less wobble and scuzz)
  • A boring font (i.e. a standard font, not a web font)

To measure the relative success of these variations, I decided to use some pretty standard goals. Bounces, page views, and session duration.


A clearer font will reduce the bounce rate, increase the session duration, and increase the page views.


If the font was totally repellent, people would ditch out on the first page.

Page Views

If the font was off-putting, people would be less likely to view further pages.

Session Duration

Some care had to be taken here. Typically a longer session duration would be good news, but as the experiment was about fonts it could indicate that it was making things harder to read. If it correlated with high bounce rates, the font is probably hard to read for everyone. If it correlated with high page views it would mean the font is reducing friction and making people want more.


Here are the results, with a sample of 15,000 sessions.

Probability to be Best (%)
Variation Bounces Page Views Session Duration
Original Font 60% 87% 60%
Clearer Font 21% 4% 23%
Boring Font 20% 9% 16%

This surprised me. My genuine expectation was that I’d have to post the following conclusion:

The result shows the boring font outperforms funky fonts. The problem is, how can I make my website look like a journal with such a boring font… as it is a personal site, surely I can justify compromising and using the clearer version of the typewriter font?

But the result clearly shows you hate the boring font. Hooray.

On the downside, I don’t really want to leave the website in a state that is making it harder for someone to read.


So my next idea is to run a mass design-by-committee exercise to find out what you think now you’ve read the background on this whole experiment. You can see the results from the Google Optimize experiment and the reasons I’m conflicted – now it is over to you.


After running the survey for a week, we ended with tie!

Font Vote Result

So here is the actual result…

The data said leave the font as it is, but I know there is at least one user who is struggling with it and I’m not happy with that situation. My decision is to update the font to the clearer version – once your caches all clear out, you should see this change.

If this still doesn’t solve the readability problem, please get in touch – ideally with real contact details so I can ask questions and learn more about how I might resolve the issue – I have another plan if people are still struggling even with the change, but I warn you in advance: it’s Comic Sans.