This article will explain how to use results from EyesDecide so that you can get the most value from the service. Please also have a quick read of the Results page, and look at some of the examples, so you can understand how the data and tools work.
What can eye tracking tell me?
EyesDecide is great for understanding which features or design elements users notice, in what order and how often. If you motivate users with a specific task via the instructions, you can also understand how they approach that task. For example, you can watch the flow of attention between buttons, pictures, wells, callouts and other design features.
History of Eye Tracking
Eye (gaze) data is a fascinating window into someone’s mind. Gaze is very closely related to attention and therefore quite often hints what someone is thinking about. Of course, people have other thoughts and impressions that are not visual and not related to gaze. You need to bear this in mind when trying to understand what a user was thinking at the time.
Gaze is useful because it provides feedback that isn’t captured by any other method. Gaze is partly conscious and deliberate, and partly unconscious and involuntary. When looking at visual media, eye-gaze reveals the process of exploration and discovery that a user follows to understand the media. Users aren’t always aware or able to vocalize why or what they look at.
Gaze behaviour is therefore affected by task motivation. An early study by Yarbus in which subjects were shown a photo and asked different questions revealed this effect dramatically (see picture). Gaze is also affected by internal biases and past experience.
Eye tracking can provide value by giving you the evidence to choose between competing designs or interaction ideas. Often, these choices get made by client preference, which may be an arbitrary choice especially when the client is not a design expert. EyesDecide can give designers the evidence to prove that a design achieves specific objectives, such as drawing attention to particular features or information. This validates the design and can quantify and justify the cost of undertaking a professional design project.
Diagnosing an Interaction Problem
Users often come to EyesDecide with a particular design problem, such as “why do users add to the basket but not checkout?” Obviously one potential reason is that they don’t know where or how to checkout.
EyesDecide can help with these problems by showing you what users actually try to do. Ask them to checkout and show them the page. See where they look for the checkout options, and then move the checkout button to where users expect it to be. If users look at the checkout button but don’t notice it, you’ll need to update that icon or button to make it clear. We have had users with these exact problems.
Rapid Design Validation
If you’ve modified a design, use the Instant Replay feature in EyesDecide to compare visual behaviour before and after. Does user behaviour now match your expectations? Sometimes the result of a modification is not as intended.
EyesDecide is intended to be used for rapid design iteration and validation on small sample sizes. Every time you try a new layout, test it on a handful of colleagues or users. In many cases this is enough to see whether people behave as expected.
You can use both the instant replay and aggregate reports (heatmaps) features in EyesDecide to discover unexpected behaviour.
The classic heatmap is an “F” shape covering the screen without scrolling. This reflects a tendency for people to begin top-left and follow text and visual cues to the right, repeating from top to bottom.
But in many cases the design of a page will eliminate the “F” shape; for example, a single centered column design will give a heatmap that looks more like an “I”. Often, features tucked up at the top in menu bars are ignored by users, who start scanning at “hero images” or other more central features. Rendering a heatmap will show whether this is the case, and Instant Replay will enable you to understand the order in which the page is typically scanned.
Once you know how users actually look at your page, you can take that knowledge and factor it into design improvements.
Identifying Behaviour Patterns
When you start to look at individual gaze tracks using Instant Replay you’ll notice that while there are variations in behaviour, users often fall into discernible groups. Some are careful and methodical and work systematically through a page, for example from top to bottom. Others explore first, zooming quickly from end-to-end and then coming back for a longer look at particular features.
At first this might seem too much to handle, but with a few more samples (10-20) you’ll start to notice the behavioural groups. Within the Instant Replay you can “tag” views as belonging to particular groups (currently limited to A, B, Invalid and None). After marking a set of views, you can use these tags to render heatmaps on these subsets to compare group behaviour. You can also check whether particular design changes will be compatible with each group.
What not to expect
EyesDecide gives you the precision to understand which features a user looks at. The heatmap may not always be exactly centered on the feature, but this doesn’t matter. Eye tracking isn’t an answer in its own right. Eye tracking is data that needs to be considered along with other sources and interpreted intelligently. Eye tracking data should not be presented standalone, but as part of a report detailing questions asked, evidence obtained and conclusions drawn.
You shouldn’t try to use EyesDecide to watch users scan line by line through dense text. There are several reasons for this:
- People don’t read by moving their eyes smoothly, especially fast readers: Instead, they jump from point to point and read a handful of words at a time. This is why newspapers have narrow columns; it’s easier for fast readers because they only need to move their eyes down, not from side to side.
- People reading websites tend to skim-read, unless they really concentrate on one paragraph of text. So you will often not see concentrated reading. Instead they will jump around.
- There may be small offsets of 1-2 cm meaning that unless quite a large font is used, there may be uncertainty over which line is being read.
- Focusing on reading lines is often a distraction from the read problem, which is the gross design or layout of a design. Reading lines says more about the user’s reading ability than it does about your content.
- A “wall of text” (like this page) is likely a bad design anyway.
Raw accuracy numbers, including precision, latency, samples-per-second etc. are provided on the technology pages. Often, results can be interpreted equally well with small offset errors, as long as it’s clear which feature they’re looking at.
Quantitative and Qualitative Analysis
A quantitative analysis produces actual numbers, sometimes absolute (e.g. “average total dwell time in area X, measured in seconds”) and sometimes comparative ( “56% of viewers looked at feature X in design variant A, compared to 72% in design variant B”). EyesDecide supports quantitative analyses – see our notes about sample size.
Qualitative analysis requires the experimenter to interpret and reason about the data. Although the data may be numeric, the analysis is about understanding and weighting factors rather than relying on the data itself to provide a conclusion. EyesDecide is the best tool available for webcam eye tracking, because the high quality results & Instant Replay feature allows study organizers to replay individual gaze tracks in detail.
Both types of analysis are enhanced by using EyesDecide in conjunction with complementary technologies. Each piece of information helps to build the complete picture.
EyesDecide is a great addition to focus group sessions. You can use it like a conventional eye-tracker: Set up a desk in a good location and capture eye-gaze as part of your session. A good study model is to replay viewers’ gaze-tracks back to them, and pause on every interesting action. Ask the viewer what they were thinking, and why they looked at that particular feature. Often, seeing the gaze replay will improve recall and detail. This type of session is called “Retrospective think-aloud”. You’re asking viewers to annotate their own gaze behaviour after it has been recorded.
Don’t forget that you can also prime or motivate viewers verbally when running focus groups, before starting them on eye tracking tasks.