TextPrefs Survey Report Part 3

Written by Jamie Freeman

User preferences for reading text on the web

Font size

Hot on the heels of type face in terms of the impact on readability is font size. A 'readable' font can quickly be rendered illegible at too small or large a size.

The following preferences are of course only relevant to those respondents who chose the particular font in the first place. Someone who found Arial to be most readable doesn't express an opinion about the preferred size for Helvetica, for example.

Arial was most popular around the 14pt to 15pt mark, split fairly evenly between about 44% of respondents, although 14pt shades it slightly. The next largest group couldn't quite decide between 12pt and 13pt, which made up just under 20% of the votes between them. 16pt and 18pt got around 7% between them, with the rest being made up of a smattering of larger sizes, with the enormous 22pt getting a couple of percent of the votes.

Helvetica showed similar results, but with a bias towards sightly smaller sizes. For this font, 14pt was less popular than 12pt and 13pt, which both gained ground to around 15% at the former's expense, which came in nearer 13%. 15pt was therefore a clearer favourite in Helvetica, although with a little over 20% it was almost identical to the result for Arial. There was not much argument to be had for one size over another anywhere from 12pt to 14pt, and there was a similar long tail to Arial's up to 22pt, with a slight peak at 18pt.

Lucida had a much more predictable 'curve' to the preferences, with a peak around 13pt (with over 20% of the votes) which tailed off either side. 11pt was about as small as people wanted to go with this type face, and the tail to larger sizes petered out at around 18pt (although a few people did like it larger) with a slight peak at 17pt. 13pt was therefore the 'winner' but if you used 12pt or 14pt you wouldn't face too much argument.

Trebuchet on the other hand is a very mixed bag, although it does have one of the clearest size preferences of all the type faces surveyed - 13pt, which garners over 30% of the votes! Either side of that size, 12pt and 14pt get around half the number of votes each, with 14pt the more popular. Then there's a 'hole' around 16pt, whereas larger sizes from 18pt to 20pt get a much stronger showing than the other fonts so far.

Verdana shows an almost bizarrely flat response all the way from 10pt (about 5%) to 18pt (just under 10%), with a couple of dips at 15pt and 17pt (to below 5%). If you're looking for a clear favourite I'm afraid our respondents are split evenly between 12pt, 13pt and 16pt, all of which are about the 12% mark.

Times has a similar response curve to Verdana -?flat and wide across the font sizes from 12pt to 20pt. It shows a clearer winner however, with 15pt standing out as the only one breaking the 10% mark with about 14% of the votes.

Georgia demonstrates much clearer trends, thankfully! A prominent peak of preferences for 14pt (just shy of 30%) starts building at 12pt (with about 12%) and trails away sharply after 15pt, which has just under 20% of the votes. A very low peak emerges around the 20pt mark, but this only accounts for a few percentage points.

Courier, the ugly-but-useful duckling, shows 12pt as the massive winner with almost 50% of the voters deeming it most readable. The type face doesn't show as much liking for larger font sizes, with no real trace of preference for sizes above 18pt, in contrast to some of the other type faces.

In summary, I think it's fair to say that your fonts are in the 'readable zone' if they're around 12pt to 15pt.

Tomorrow things get really interesting when we introduce column widths into the mix!

< Read Part 2 | Read Part 4 >


  • Martin Tóth -


    These numbers are great, but could you post some "curve graphs" also? (Or some overall numbers in spreadsheets?)

    Or do you plan to share some documents after all survey blog posts are up?


  • Kate Naylor -

    Excellent stuff, Jamie. I'm always banging on about clear formatting in my copywriting, marketing and SEO blog.

    Would you mind if I featured your research in my blog? If so, would you like editorial approval before I upload the post?

    No probs if not… but every little bit of link juice helps!

    ;-) Kate

  • Mads -

    I agree with Martin, these numbers are interesting, and luckily, they're not too surprising either. But some line graphs would be nice to see for each font, or even better, the underlying data. Are you planning on releasing the data in the end, so the community can fiddle with it and try to analyze it themselves? Some clever guys may even come up with new interpretations of your data!

  • Jamie Freeman -

    Hey Martin, tune in tomorrow for some graph goodness!

    Kate; we've made our research public, so please feel free to blog about it. Of course, we want people to know it's our research so appropriate credits and links would be much appreciated :-)

    Mads; not sure about releasing the raw data itself, but we've built a load of graphing and analysis tools which we may well make available. Watch this space.

  • Gary Miller -

    Great stuff!

    A lot of it's almost intuitive, but it's always nice to have some qualitative data to back you up.

    I agree with others - graphs please!

    Cheers Jamie!

  • Ian Eiloart -

    This is all very silly. why not let me choose my own font size?! The Internet is not paper!

  • Joe Holdcroft -


    All modern web browsers have the capability for you to resize text to your preference. Most even include full-page resize now. What more can you ask for? :)

    Creating controls for text resizing should not happen on the web page - this should be handled by the system (browser/OS).

    This research and it's findings are really just guidelines for default behaviour.

  • Florent V. -

    Thanks for sharing these results. But i’m puzzled by the use of points. What you tested was on-screen readability, using web technologies, right? So are your results in this post using REAL points or FAKE points?

    There's a lot of misunderstanding with the pt unit on the Web. So let me clarify this distinction between real and fake points:

    1. Using real points, an object whose size is 10pt will have exactly the same physical size (with maybe up to one pixel of difference because of rounding), whatever the OS, screen and configured resolution. This means that the amount of pixels used to draw the 10pt object will vary depending on those parameters.

    2. Using fake points happens when a web browser decides that 1pt should be translated to 1.333px (which is correct only if the screen has a true resolution of 96dpi), or when the browser uses a DPI value returned by the OS and that value is incorrect (e.g. Windows and recent OS X use a conventional value of 96dpi, which is almost almost always wrong).

    Could you clarify what you used when testing? Did you use CSS to set a `font-size:Xpt` or something similar?

  • Jamie Freeman -

    Hi Florent. Fonts are commonly referred to in terms of point sizes, and as I've been a designer for over 20 years (and my father one before me) it's simply the way I talk about fonts! You can of course express them in a number of ways in terms of CSS. I would disagree that there's a lot of misunderstanding about it actually. I know what a point really is, but I know that on the web it means something different, depending on screen resolution. (I also know that when young people say 'sick' they really mean 'good' ;-) )

    Your explanation fails to say that a point is 1/72 of an inch, which of course made sense when computers ran screens at (or near) 72dpi. Things have changed, but I think most people reading this level of detailed research would be aware of the size of a true point. Those newer to the game might tend to simply think a point is a pixel, so either way I don't think it will cause much confusion.

Any questions? Please get in touch…

We'd love to talk to you about your web project, e-commerce, EPOS system or bespoke web application.

Message Digital Design Ltd

+44 (0)1273 772247

© Message Digital Design Ltd. 2018