My quick answer: There isn’t one…so you shouldn’t try too hard to make one.
If it’s an icon, then it’s worth up to three words—at best! The oft-cited cliché is very misleading because icons are a very poor way to communicate. With the exception of well known standard icons, people understand text labels much faster than icons.
The ribbon UI, introduced by Microsoft Office 12, is much easier to understand and use than traditional toolbars primarily because it takes extra space to give nearly every command an explicit label. (The exceptions in Word: Fonts, Paragraph, Quick launch are unlabeled.) Office also uses many preview-based graphics (such as Styles that preview the effect of the style), but those are really thumbnails not icons.
Don’t believe that icons are a poor way to communicate? Take the “icon challenge” by removing all command labels and seeing if you can correctly determine what the commands do based on the icon alone. For example, try to figure out the labels for the Insert tab in Word:
When I tried this, I scored only 10 out of 24 for this ribbon tab. Keep in mind that Word is a familiar program and Office uses excellent iconography, so I would expect the typical score to be even lower.
BTW: I recommended using this approach to evaluate your product’s icons.
If icons are so poor at communicating, why bother with them? First, I should reiterate that well known standard icons communicate their 1 – 3 words quite effectively. It’s the not well known, non-standard icons—the ones that require time and thought to figure out—that are the ones in question.
Such icons have value not because they communicate their purpose well, but because they help users recognize and distinguish commands visually. It’s all about efficient visual recognition. So while users understand text labels quickly, they can recognize and distinguish between icons faster still. For example, users might remember that the command they are looking for has a globe on it and locate it immediately, even though they might not know what the globe means. When there are many commands (as on a typical ribbon), the icon + label combination works well because the icon enables quick visual recognition and the text label enables quick comprehension.
Efficient recognition is extremely valuable—just keep in mind that it’s no substitute for comprehension. If your target user’s comprehension of your icons is low, it’s likely that you need to reconsider your labeling strategy more than the icon design itself.
The user’s ability to understand an icon is primarily determined by the icon type. The follow icon types are usually easy to understand:
Standard and simple works well. This list reveals an interesting challenge to icon design: Icons are pictures, and pictures show nouns. Yet, icons are used to represent commands, and commands are usually verbs. Consequently, most icons boil down to a noun representing or doing the verb.
The following icon types are moderately difficult to understand:
Metonyms and synecdoche are related to metaphors, but I listed them explicitly because, contrary to popular belief, metaphors aren’t the only game in town. Using a fork on a map to represent a restaurant is a synecdoche, not a metaphor. Again, simplicity and familiarity is the key to success here. For example, a star is a successful metonym for “favorite” because people often rate things they like using stars.
The following icon types are difficult to understand:
Going back to my original question, “fidelity” is an abstract concept, so it’s very difficult (I would argue, impossible) to create an understandable icon to represent it. One could try: Dogs are known for having fidelity to their masters, but a dog icon is far more likely to be interpreted literally.
In addition to type, context plays an important role by allowing users to easily deduce meaning. For example, a zebra icon (an unfamiliar noun, icon-wise) is meaningless out of context, but in the context of monkey, turtle, bird, and snake icons, a zebra most likely represents savanna animals.
I have a couple laws for icon design:
The longer it takes to come up with an idea for an icon, the less comprehensible the icon is going to be.
If an icon requires a tooltip to understand, it’s not comprehensible. At best, using it helps recognition.
If you’re wracking your brain trying to come up with an idea for a good icon, most likely it’s because there isn’t one. Once you’ve made this realization (and you really must have an icon), better to focus on the recognition consolation prize.
Consider the following, in priority order:
Given the challenge and expense of creating meaningful icons, it’s important to reuse icons whenever appropriate (as opposed to creating new ones). To reinforce their meaning (instead of diluting it), choose preexisting icons based on their meaning, not their appearance. If a design detail has a different meaning, use another design. So, use scissors to mean Cut, not Office supplies; use binoculars to mean Find, not Zoom; use a gleam overlay to mean New, not Glossy.
To show how inconsistent reuse can dilute meaning, consider the ‘x’ overlay. Currently, there’s no consistency at all, either in terms of meaning (does it mean delete, error, cancel, close, exit, stop, clear, disconnected, or not available?) or presentation (red, black, or while; normal vs. script; alone vs. within a circle). Consequently, you can’t be sure of its meaning based on the design alone—all you know is that the overlay indicates a state that isn’t positive. (BTW: The design community should fix this: Delete should always be a black, script x (never red!); Error should always be a red, normal x; etc.)
Bottom line: To preserve meaning, these design details aren’t arbitrary choices.
If you do only one thing:
Reconsider the need for icons. While icons are helpful for comprehension and recognition, users’ ability to comprehend icons is vastly overrated. Consequently, prefer icons that are standard, simple, and familiar. With the exception of well-known, standard icons, label all icons either in-place or with a tooltip. If you really need custom icons, use an icon design specialist.