Undercover User Experience

Notes from Cennydd Bowles‘ presentation at London Web on Thursday 15 July 2010. Cennydd is currently at Clearleft.

The User Experience (UX) disease

Some businesses get user experience, but for every one that does there are 999 that don’t. Companies like that aren’t interested in serving the needs of their users.

In the past Cennydd’s had to make companies take usability seriously. Now, undercover user experience is designed to help usability professionals to progress in these companies.

The Undercover Usability Manifesto

• Go undercover
• Ground up, not top down – don’t approach your boss asking for resources, just do it.
• Change through small victories
• Delivery, not deliverables
• Good today is better than great next year
• Work with people, not against them
• Action, not words

Let’s take action. Two big things you can do now:

1. Expert Review

Using heuristic evaluation. Step through processes with your website, is it:

• Made for humans – is it relevant, useful or enjoyable? Does it fit their mental model?
• Forgiving – does the site prevent errors, does it minimise them?
• Accessible
• Self evident – is it clear what the site’s for?
• Predictable – does it use known web conventions
• Efficient – is the site responsive
• Accurate

2. Analytics

This is a topic that’s not given enough attention in user experience. Looking at the stats, consider:

• Visits per unique user – are people coming back?
• Entry pages
• Bounce rate
• Referrers
• Keywords
• Navigation paths
• Registration / purchase conversions – where are the largest drop offs?
• On-site search – an underused trick. A real gold mine of understanding what the user is looking for. Test those search terms, are they finding what they want?

The real user experience process takes time. Do your business research first. It’s tempting to jump in without finding out:

• Objectives of the project
• Requirements
• Un-requirements
• Exclusions – things that cost too much or services like Flickr have sewn up

You can never find all this stuff out because it’s held in tacit knowledge. If you can never know everything, you are allowed to get it wrong. This is a relief, you don’t have to get it right first time.

The best way to be a user experience designer is to understand the culture of the company. Several warning signs exist.

1. Cash cows. Most companies have them. You won’t be able to make changes to these unless you can prove you’re going to make the company more money. These are dangerous because changes may go wrong and you might lose the company’s interest in usability. Be careful.
2. Paralysing process. It’s difficult to get new processes and new techniques into these companies. It’s only when you come up for air, that’s when people will challenge your work. Don’t be overly eager to share your work until it’s ready. If results are already positive that will skew the company’s view in your favour.

OK, real UX

You can do some quick and dirty qualitative research. Remote research is becoming a big topic in user experience. Phone your users up, yes you’ll miss some nuances, but at least you’ll get something. Speaking to your customer service staff is also useful, their insights can help you understand who’s coming to your site.

Quantitative research is also easy.

Finally, market research can give you information on psychographics and demographics. Data driven personas help you design for a particular type of user. Data driven personas don’t require much one-to-one research with users. It’s not scientifically valid, but if you treat this as a living document it’s still much easier to design for this ’person’ than a collection of numbers.

If you’ve done your research, make some small changes and see if they work, if they do you’ll start to be included in decision making discussions.

At Clearleft they use design games like divide the dollar (which I think is this) to come up with ideas.

The undercover UX designer needs to stay as low-fi as possible (e.g. pen and paper). It doesn’t matter on the tool, you need to convey as much information as possible in the smallest amount on time.

Test informally. Use a laptop and a cafe. They use Silverback (since they made it). Get people to talk aloud and explain any difficulties they have. You can even do remote testing. The videos and the output you get can be very persuasive.

You need to make sure the business is happy with the work that the user is happy with. As designers, UX people get much more then developers because everyone has an opinion. To run a quick critique session, get the stakeholders in the room. Explain the decisions you’ve made before you show them your designs. The real trick is to not get into an endless loop of changing things.

The idea is to list what needs to change to meet the clients approval and the users approval. Then get these changes signed off.

Don’t make user experience combative.

There are three tiers he uses when discussion usability decisions with clients:

1. Direct feedback from the user, if he doesn’t have that he’ll go back to the…
2. Research he did at the start. If he has neither of these he’ll fall back just on…
3. General points of good design.

Getting to know people (and beer) help the process.

Q and A

I didn’t really take notes on this except he mentioned a new service called Performable just out of beta, worth checking out. Of course you can use Google Website Optimiser as well. Of course, this presentation was also a lead up to his own book, Undercover User Experience.

He also mentioned a book by Dan Saffer, Designing for Interaction.

Posted on Friday 16 July 2010.

Posted in books, notes, usability | 1 comment »