Great article. It’s a great way to demystify types of images from a usability perspective.

However, as some of the others mentioned it would be useful to have some stats on these image categories and their uses along with type of impact on different segments of users. (Ornamental to one maybe content to another and vice versa.)

Does accessibility play a part as well?

Branding has a big impact on the type of images used (any marketer would swear by it). Though I agree, overdoing it will be more harmful.

I think including stats on international Web sites where applicable would be useful in order to be fair to an international audience consuming this information.

Finally, (not least from an accessibility viewpoint) I think that any site that has to display a picture of its products instead of a description has a big problem – there is no bigger turn off for me than a site that cannot describe its own products.

The images of people shaking hands came from a stock photo house absolutely. Maybe the company did not have the budget to take a real photo shot to marry the text content into an image that Citi could actually own. Maybe the site team could not understand the need to create an image that represented the text to bring it to life. Maybe they lacked professional design skills. Or maybe they just did not care. Who knows. The art in creating images is to give the flat words life. The web is a visual medium. People skim through text and images and in a nano second need to understand the page context. If done properly text and images can lock in the understanding.

Often times creative teams are at odds with branding. Branding will come up with a celebrity spokes persona and then say plaster this every where. Mistake. Use the personal to set up the brand position and then get rid of it and let the site guide the user through whatever they came to do.

Neither an art nor a science, “graphic” design adds life to sites but has to be done with the skill of a water colorist not a house painter. I offer this site as a fantastic example that marries imagery with content: http://www.arcteryx.com

If you change the conversation to visual design, you include layout, typography, color, imagery, interaction design, animation, etc. In this case all of the visual elements work together to communicate with the user about what to do. Simple uses of color and layout can help users know where to start and how to use a Web site or piece of software. Similarly they can be used to communicate brand or elicit a response. In any event they serve a purpose.

However, if you only talk about graphics, you miss the point. Any page or screen element should contribute to a conversation with the user about what they need to do or how the company wants to communicate with them. The rest gets back to “gratuitous graphics.”

http://www.surl.org/usabilitynews/112/aesthetic.asp

http://timesonline.typepad.com/science/2009/12/climate-sceptics-get-it-wrong-1.html

and

http://timesonline.typepad.com/science/2009/12/climate-sceptics-still-getting-it-wrong.html

“If we observed the user gleaning meaning from it”

I think the question of the hour is which images provide NO meaning to a user? Even the aforementioned Carmen Electra provides a sense of celebrity taste – inherent notions of trust and belonging.

I find categories such as Navigation and Content very useful, but I have a hard time assuming such an expansive category as Ornamental covers “everything else”. Even arbitrarily-placed images can create meaning with a user – aren’t there degrees between subconscious association and purple chairs?

As @Juan points out, different kinds of content have different meanings based on their context. In other words, maybe a designer should focus less on what’s supposedly empirically and universally “useful”, and figure out what’s contextually relevant.

I submit that human beings don’t subdivide their experiences into pieces like “graphics” and “content”, and any designer worth their lunch money doesn’t design that way either.

Or, as Frank Lloyd Wright put it, “form and function are one.”

I have run split tests where putting a person on the page with their gaze pointing toward the call to action has had a positive impact on page performance.

I would love to see test results to confirm the value of ornamental graphics in your study. Great work overall, and thank you very much.

Also, is there a type of imagery that can play a purely visual (i.e. not semantic) role by making the page easier to scan by breaking up text or alternating patterns, etc…? I realize that this is a difficult contribution to assess (except using eyetracking, of course but we do know that page layouts that are overly text-heavy can be off-putting to many users. So, in this sense, maybe imagery can act in a way that could be similar to non-image graphics (lines, backgrounds, etc.).

We’ve been looking at emotive qualities and branding qualities for quite a while now. We measure people’s emotive states before and after the sessions. We also measure their brand engagement before and after the sessions.

Over the years, we’ve found many factors that increase emotions, such as happiness, empathy, and even shame. We’ve also found several concrete methods of increasing brand engagement through design.

However, the use of ornamental graphics doesn’t do either.

Now, the sad-looking puppy, you mention, depending on the image and the context, we might not categorize as an ornamental graphic. If we observed the user gleaning meaning from it, it becomes a content graphic. And we do see content graphics having the effects you suggest.

I’m not buying the same-image-on-all-the-pages strategy. If the goal is to lighten the text on the page, by eating up the real estate, why not just get rid of the image and use the real estate more judiciously?

Show me where it’s written that pages need to be rich. If the image isn’t serving the needs of the user, it’s distracting. To have a random image on the page requires some sort of cognitive decision to pay attention or ignore. Why add the load?

Of course, maybe you’re fixing the wrong problem. Instead of putting graphics on the page that are meaningless, because people need something other than the text too look at, why not rewrite the text to be more compelling, thereby not needing any random image support? Or find content graphics that support the text?