Of all of the different tools we have to evaluate the effectiveness of a website, eyetracking is probably the most misunderstood and underutilized.
Eyetracking is the mechanism we use to observe and measure what people actually look at on a page. For individual users, they show what the order in which various elements caught their attention and how long they looked at a specific area.
Figure 1: This example from the Universtity of Minnesota shows in what order and for how long a single user looked at different objects on this page.
Typically they also produce heat maps that demonstrate what everyone in the test looked at while they were on the page.
Figure 2: This example shows the areas where a group of users fixated their gaze most often (shown in red).
What people look at and consequently do is directly tied to what they came to the page to do in the first place. However, people don’t keep seeking until they are 100% certain they found the right answer. They typically stop at the first reasonably plausible one.
In general, the better your design is able to both 1) draw peoples’ attention to the area of the page that will enable them to complete their task and 2) enable them to recognize it as the solution to their problem, the better the page will perform at helping them accomplish their goals and make them happy customers.
So how do you know for certain that that they’re looking where you want them to look so that you can improve your design?
You could ask them what they look at. User interviews, focus groups and properly constructed surveys will help discover what people think. But for more reasons than can be discussed here, they are inadequate measures of behavior.
Site analytics measure behavior, but they don’t explain why that behavior occurred.
Professional usability lab studies explains the “why” question to a great degree – and I recommend at least a quick, inexpensive informal study (often several) for most projects – but it still requires interpretation. Eyetracking can help the design team understand, contextualize, visualize and interpret these problems.
How to Get Started
There are 3 general approaches to getting started with eye-tracking studies. Which one you choose depends on your goals, expertise, capacity, deadlines, margin for error and budget.
- Outsource
Outsourcing is the ideal option for organizations looking to manage the performance of a high-value project that eyetracking can measurably help improve, if your team does not have equipment, time or expertise to conduct them. - Bring it In-House
Organizations that regularly build and refine websites should consider whether training and hiring in-house experts makes sense. The equipment and software can be obtained at a fixed cost (systems sell for over $20k), and in many cases incremental costs can be relatively low. - Simulate
If it could be shown that the human eye is typically drawn to certain objects with defined characteristics in specific contexts, accurately predicting what people would look at on a page is a matter of putting the right algorithm in a screen interpretation engine. Eyetracking simulation engine AttentionWizard is said to be able to do just that for static images. Its creators claim 75% correlation with actual eyetracking studies. This is partially because it cannot interpret the context of a specific task, but with prices for testing each image far lower than traditional studies, accuracy at that level may be worthwhile.
Limitations
Eyetracking doesn’t measure peripheral vision. Just because someone’s eyes did not fixate on an area, it doesn’t mean they didn’t see it or weren’t affected by it. Nor do they read minds. Just because it appears as though someone looked at something, it doesn’t mean they comprehended it.
Figure 3: This example from useit.com shows that users looking for the current U.S. population looked and fixated directly at the number that answered their question. However, as Jakob Nielsen reports, only 14% of users successfully identified it for what it was. A closer look shows that users fixated on only the left part of the number, implying that they did not really comprehend what they had seen.
To answer questions you have about a Web product (be it a site, application, or some hybrid of both), make sure that the test actually measures what it is you need to study. Just as you would not use a stopwatch to measure the temperature in Hawaii, nor should you ask analytics, surveys and usability lab studies to measure what areas of a page draw peoples’ gaze.
No single research method measures everything, but in many circumstances eyetracking can be a good supplemental tool in your design arsenal.
Related Links
Eyetracking: Is it Worth It?
First 2 Words: A Signal for the Scanning Eye
No Comments | Trackback | Categories: Testing, Tips and Tricks | Email This Post
Leave a Reply