Well, if I were a company designing an information service, perhaps monetized by ads, then it could make sense to collect eyetracking data to see what people tend to look at, perhaps as part of an AB testing regime.

Then the choice comes between: do I want to bring people in to my expensive lab where I can get high-fidelity results, or do I want to potentially get a much larger sample for cheaper at the cost of lower fidelity.