Using Website Session Recordings To Identify Features To Build
As a founder, I managed the product at Fidoc, a doctor booking platform. While poring through the analytics, I came across a curious yet annoying stat. Almost 40% of the users who landed on the homepage would exit the website without visiting any further pages. And while the bounce rate was pretty disappointing for the homepage, the other pages on the website fared way better.
Based on numbers alone, it was difficult to figure out why this particular user behavior occurred. I needed something more - I remember telling my developer "I wish there were a way to actually SEE how your users were interacting with the website".
So I looked online for a couple of days, and came across a product called SmartLook. This tool would essentially provide you a video recording of each user session. I couldn't quite believe it yet, but I got it implemented quickly, and it was exactly what I'd needed! Suddenly, I was looking at how users navigated the home page - they would scroll from point A to point B, pause for a few seconds, scroll again, and then click on a link or exit the page.
Within no time, it became pretty clear why the bounce rate was so high - our search box. Half the users simply could not find the result they were looking for. Specifically, I mapped out 3 issues with the search box:
1. Symptoms instead of speciality
If they had a headache, the user would type in "headache" in the search field, rather than "General Practitioner".
2. Incorrect spelling of speciality
Many specialities were spelled incorrectly by the user (understandable, since they are usually complex words). "Ophthalmologist" would be spelt as "Opthalmologist", and our search box would show no results.
3. Clinic name with the area
If you wanted to visit Aster Clinic in Business Bay, many users would search for "aster business bay". I suspected this was the case as many people are habituated with free search on Google. However, the default Magento search is simply not equipped for free search, and again there would be no results.
Going back to the drawing board, we redesigned the search box from scratch. We included over 600 symptoms that were now searchable, such as "headache treatment" and "wisdom tooth removal", and simply linked them to the right speciality. The most common spelling errors were included in the backend, which would identify and deal with the errors. Finally, we made permutations of clinic names and localities that sort of mimicked free search. For example, "aster business bay", "aster clinic in business bay", "aster in business bay" were all pre-loaded search terms.
The changes were pushed to live, and within a week, the homepage bounce rate was brought down to 25%.
Here's a great article that summarizes session recording and how you should implement it.