Many times it is difficult to see where the feedback is used in the GDD methodology. In the end, in GDD the collection of statistics is key in order to make the web work in the best possible way, but it is very diffuse the type of changes that the GDD allows facing your web, so I will put three practical examples.
The first example is a web page that had the home page divided into 6 groups, each with the name of the 6 services of the company in text. It worked the wrong way, because the user did not know very well where to click. They were reduced to 4 sections with icons and the thing improved drastically, disappearing the problems previous to that situation. In itself it was seen that the web did not work well for reasons yet to be determined and we proceeded to realize a hypothesis of improvement that finally was correct (as it could not have been). The users did not arrive at the beginning to see the products (or they did in a very small percentage) because they were frustrated and they went out. With the improvement it was possible to quadruple the percentage of users who passed the initial web.
In addition, sometimes during the development of the webs mistakes are made. This is something that can not be avoided because knowing how the behavior of the user is going to be at all times is an impossible task. So, the ideal is always to perform an analysis of user feedback and improve the usability of the web. This is the case, for example, of texts that contain terminology that is difficult to understand because it is too technical or due to a specific geographical area (Andorra goes a long way with certain words in Catalan). One case that I remember is that of a customer who had a product called in a technical way (in English) and his clients looked for him in Spanish in a more colloquial way. Then, apart from being difficult to position, the people who came to him did not identify him as his solution, a disaster. It could be corrected and started to have more visits and to work better.
Another specific case that I remember is that of a customer who put a long text of a particular product, leaving in general a page that looked more like an article than the presentation of a product. The users were confused, because in addition the rest of pages of products were of another very different form. Separators were created between the text with images and part of it was placed more colloquially, which caused us to move from 3% of complete readings to more than 15% (which in this case is great).
There is a very long article for the format of the blog, but there are a lot of cases similar to these. The GDD makes the web work better and in the end that is noticed in the results.