Manual software testing is easier and cheaper than automation, but also, it's time-consuming and reduces testers' efficiency. However, sometimes you can't deal without manual testing during cross-browser testing. Such cases cover areas when automation can't replace humans' minds and perceptions.
Let's consider cases when manual testing is necessary.
To reveal unobvious failures. Finding some failures depends on the experience of testers and the knowledge of the target system and browser. Also, some bugs can happen under specific conditions that automation testing can't cover. Through exploratory testing, which is always manual, testers can find atypical bugs and problems.
Check a visual environment. Automation can cope with checking the correct location of visual elements. But testing of app appearance, smoothness of animations, and overall usability it's work for humans. Thus, only through manual testing, you can ensure how animations work or how components of design look in different browsers and under different conditions. HTML5 and CSS3 open new opportunities and allow developers to create new effects and elements. Moreover, effects can be displayed even if JavaScript is disabled. But as HTML5 and CSS3 aren't standard, their abilities can be displayed in some browsers incorrectly.
Check UI. Design components must not only look good but respond correctly. With manual functional testing, testers can check how different fields, buttons and forms behave on different browsers.
