90 percent of everything : Usability Blog

Written by Harry Brignull

Return on Investment (ROI) analysis is normally thought of as something that only businesses do in trying to work out how to make cost-effective decisions. In fact, ROI is something that normal consumers think about a great deal, albeit in an informal capacity. In fact, understanding how consumers make these judgement calls is very useful to know.

When business think of ROI, they think of the investment and return in financial terms. This is convenient because they are comparing like for like, so all the data can be pumped through spreadsheets and models. Consumers, on the other hand, don’t have it so easy. Their investment is made up of perceived time and effort, as well as cash payment. And they don’t know how much time and effort is required until after the act, so they are constantly having to keep track of the “feel” or “scent” of an experience.

Throughout the entire duration of an experience, users will weigh up the perceived effort against the perceived reward. This is a balancing act: so long as they believe the reward will outweigh the effort, your site gets the green light. As soon as the balance tips towards more effort than reward, they hit a tipping point: a “screw this!” moment, when they decide it’s not worth their bother and give up.

What’s interesting here is the scarcity of the information that consumers use to make their judgements. Sometimes a site just doesn’t feel credible, and this can be down to a cheap logo or dodgy typography. Other times, if you have a little tussle with a tricky form, it’s enough to make you give up and leave. It’s the same when you buy a second hand car. You look for rust in the wheel arches, listen for odd sounds when test driving, but ultimately you have to make generalisation, based on a few sparse facts – i.e. “Based on what I’ve experienced so far, is it worth seeing this through?”

This is why designers have to be obsessive perfectionists. Small flaws in a UI are the equivalent of rust-spots, squeaks or scratches on a second-hand car. They tell a story for what the user can expect in the future.

If you’re in South East England and into User Experience, this post will be of interest to you.

UX Brighton is a free monthly event for User Experience types. It was pretty successful last year, having speakers such as Andy Budd (Clearleft), James Page (Fera Labs), Pete Gale (Cogapp), and yours truly.

UX Brighton went a bit quiet over the past few months, but we’ve breathed some new life back into it and from now on it will happen on the second Tuesday of every month. A typical UX Brighton event involves a 1 hour talk with about 30-40 friendly attendees. After the talk we all go to the pub for a good natter, and everyone’s welcome.

Here’s some details about the next event:

UX Brighton: Mental Models, Service Design & Problems with Convergence

When: Tuesday April 14, 2009 at 6:45pm
Where: The Werks, 45 Church Road, Hove, UK, BN3 2BE Info:

Speakers: Filip Healy & GiGi Demming from Amberlight. Amberlight is a London based User Experience Consultancy. Clients include Apple, Microsoft, Sony and O2.

Summary: It’s becoming increasingly popular for large companies to bundle together a range of services such as broadband, mobile and TV. While this can result in lower prices for customers, how does it really impact the customer experience? Amberlight has engaged in a number of service design projects that have looked at this very problem – how can these bundled but disparate services be aligned in a manner that presents a coherent, ‘joined up’ feel for the customer? In this presentation, Amberlight will talk about their learnings, and will lay bare some of the key questions that you need to think about when dealing with this tricky service design problem.

→ Book your free ticket here

Mailchimp (the email marketing webapp people) have written about a really interesting piece of user research they ran using Amazon’s Mechanical Turk.

Within the Mailchimp webapp, you can upload an image as the banner for your email templates. To make it super easy to build your own template, Mailchimp are creating a new feature that parses the colour information your image and suggests an appropriate colour palette to go with it (i.e. for the page background, header, body text, links, etc). Neat idea.

So, they started out with 700,000 alternatives, which narrowed down to just under 26,000 using a few rules. However, this list was still too way long and contained quite a few ugly combinations of colours – what they really wanted was to offer a handful of aesthetically pleasing colour palettes for a given image.

So, they gathered human feedback via Amazon’s Mechanical Turk, getting 503 people to carry out 85,000 small tasks in total, as pictured below:

After some hefty number crunching, they got the number down from 26,000 to 600 palettes, of which only a small subset would be offered to a user depending on the colours of their header image. Mission accomplised! It’s worth reading their original article, since it contains a few other quality insights, e.g. how they filtered out the dud data from turkers who gamed the system.

At 2 cents per task, I think this means the total spend was $1,700 USD for recruitment and incentives. As a point of comparison, this is how much it can cost to get just 8 users (that’s not a typo) recruited and paid for face-to-face research sessions. These two types of research are suited to very different contexts and findings, so obviously this is comparing apples to oranges, but still, it makes you realise how incredibly cheap it can be to gather human feedback using Mechanical Turk.

This strikes me as an effective use of Mechanical Turk, because they didn’t need to access their own specific user base – a large, fairly random assortment of humans were perfectly appropriate for making aesthetic judgements on colour palettes. It makes me wonder what other ways the UX community can take advantage of Mechanical Turk. Any ideas anyone?

I really like Mark James’s Silk Icons. Not only are they completely free, but they are pleasingly neutral and sit well on almost any page.

At Madgex, our design team has been using them in our Axure wireframes. The only trouble was that the full colour icons were too eye catching, and the 16×16 pixel size was awkward to snap onto the page grids.

So, I’ve batch converted them into greyscale, with a 20×20 canvas size. No big deal really, but I thought I’d post them here to save you the trouble. They are redistributed on the Creative Commons Attribution 2.5 License. This means you may use it for any purpose, and make any changes you like so long as you include a link back to famfamfam.com in your credits.

→ Download Greyscale Silk Icon Pack

Amendment – if these icons strike you as too detailed and polished for your wireframing needs, then check out the Konigi Wireframe Icons by Michael Angeles. Only $10 a pop.

Here’s a short screencast demo of our new Lazy Registration system on the Madgex Job Board Platform. If you’ve read Luke Wrobleski’s book or ALA article ‘Sign up forms must die‘, this will probably be familiar territory, but if you’re not 100% up to speed with the concept of Lazy Reg, you might find this interesting.

Oh, and bear in mind the website (bigworkbag.com) used in the demo is for demo purposes only. The jobs are all fictional.

Intended for web developers with an intermediate to advanced knowledge of XHTML, CSS, Javascript and Windows 95, the script BSOD.js provides an easy-to-use class to boost the error reporting user experience of your websites. Made by Guillermo Rauch of Devthought.

Benefits include:

• Simple syntax.
• Works with every layout and browser.
• Usability tests show extraordinary levels of acceptance.
• Supports events to extend its functionality

→ Read more and try the demo here.

Mwahahaha! Seriously though, there is a serious message in here about why error messages are often neglected, i.e. developer convenience. Thanks to Bruce Boughton for the link!