How people really browse

Let's observe how people really use the Internet. And let's collect theories and models on that. When the observations don't fit the model we know we need to improve our models.

Observations

Let's play anthropologist: collect observations of online behaviour. It may help us with building better models of online behaviour.

Pattern: Social browsing
Random thoughts, April 2001.
People don't always browse alone, but we tend to test them alone. It seems to me we need more in depth research on the users' experience, including social browsing. This article is a great example of research that really goes into depth examining how users experience websites. (It also talks about the message they receive versus the message that was intended.)

Pattern: avoiding official looking URL's
Self observation, March 2001
I was looking for Harry Potter websites (a famous fictional book character) and noted I was avoiding links that looked like corporations in the search result page (I did a search on Google) . (e.g. links like harrypotter.com) I checked them out quickly, but when I had a glanca at them (I'm on a fast connection) and they confirmed my suspicion of being not driven by the Force (of fandom) I quickly abandoned them.

Pattern: Perceived security
User testing, February 2001
We were doing paper prototyping of a website which had the typical "shopping with us is 100% secure" bit on it. The user commented: "It says it's secure, so that's ok", and moved swiftly on tt filling in her credit card details. The power of language!

Pattern: URL's in search boxes
Log analysis
People still very often enter a URL in a searchbox it seems, on sites like Yahoo (e.g. portals). On a related note, I have a question box on one of my sites, clearly labeled and not looking like a search box at all, but people even use that often te enter search queries. (never URL's though)

Models

These are some models, opinions and conclusions that try to explain how people use the Internet.

Critical incident analysis

Jakob wrote about another fascinating Parc Xerox research, using the critical incident analysis method. This can help us understand why people go on the web. Turns out that for the real important things, the reasons to go on the web are:

compare information: 51%
aquire information: find out something specific: 25%
understand a topic: 24%
Satisficing.

People pick the first navigation choice that seems reasonable to them, not the best one.

We already knew people don't read webpages, they scan them. But they don't look at the navigation either!

They don't examine the page to look at all possible solutions, and then pick the best one. They don't even consider more than 1 solution and weigh them against each other. They just scan the page until they happen upon something that seems somewhat related to what they want and click on it.

This relates to a theory called "satisficing". It's used in decision making theory, and it means picking a satisfactory choice, as opposed to picking the best choice. The more classic theory is optimising: evaluating the options and picking the best one. It turns out that's not how we make decisions.

(a)They will decide on what seems useful by how it looks. E.g. if it looks like a searchbox they'll enter a searchword in it and hit submit even if it is labeled as something else than a search box. If it looks like a banner they'll ignore it.

Link scent.

Link scent is the idea that a link should give a really good idea of where it's going, by the description, what's shown around it, and so on.

The idea is people don't really mind following lots of links, they mind links that are not clear and don't deliver what they promised (or seemed to promise). Link scent is related to information foraging theory.

People stick with what they know.

Once someone has found a way to find something, and they want to find it again they'll use that same way. Even if there are better ways to find it.

Example: you would be surprised by how many people actually type URL's in the search bar of YAHOO, and then click on the result to go the website. Again and again! I saw my mom do this, and I was kindof surprised. So I talked to some people, and checked my server logs, and there it was. It turns out a lot of people do that.

Another example: have you ever went to a website because you knew there was a link to another website that you wanted to go to on it? Think about it.

Social navigation.

This is quite a cool way of thinking about navigation, it kind of ties in with collaborative filtering and such. The idea is that people are social creatures and that influences how we build navigation.

Examples are "people who came to this page tend to go to one of the following three pages: ...". (group interaction) Or: "people who liked this also like ..." (recommendations) "leave a note" (annotated pages) and so on.

Information foraging theory.

This one tries to explain how people browse the web by comparing it to how people in tribes (or animals) forage for food. Food is distributed in patches, with more empty areas between them. Foragers will typically stay in the same "patch", looking for food, until it's not worth it anymore, and then move to a different patch.

Xerox did a lot of work on this, including mathematical models that predict browsing patterns, and I may not explain it very well here but it's actually a very useful theory.