We all have problems, which can certainly be annoying, but they aren't always a bad thing. Problems, especially when talking about website optimization, can be opportunities for improvement. And that’s exactly what our Strategy & Planning department professes: site optimization projects should always start with a problem. You may have encountered someone who or maybe you yourself have read a list of tests ideas and immediately wanted to run a battery of random tests, and I assure you, this isn’t optimal. Firstly, you might solve some cosmetic glitch but you won’t likely solve the underlying issue. Secondly, websites, consumers, industries, etc. are all a different, so whatever solution worked for one company might not (and likely won’t) work for you. Starting with a problem works, and it is done by discovering the weakest points on your site, prioritizing which are the most critical, fully developing tests around those weak points, optimizing that area until it is no longer the weakest, and moving to the next weakest point. Well, how do you identify your site’s weak areas? Again, it’s not from opinion, suggestion, or a hunch, it’s from data. Use a combination of site analytics and user testing to determine your problems areas and their severity. Site analytics alone won’t uncover everything. Data, qualitative and quantitative together, will fully uncover your weak points. Proper analysis will likely unearth a great number of problems, but you won’t be able to test everything, certainly not all at once, so next you’ll need to prioritize your possibilities. There are several ways to do this. One might be finding tests that have a balance between the highest impact and the lowest difficulty. Whichever problem floats to the top, start with that one. Now that you've identified your site’s problem areas and sorted them by importance, its best to fully conceptualize the other necessary parts of a test, such as KPI (key performance indicator) and hypothesis/es. Thankfully, this is easily done when starting with the problem--those metrics are inherently contained in the problem that you’re solving for. Finally, with the problems started and sorted and measures of success developed, start testing. Keep in mind too, that tests themselves might have weak areas. A good experimenter will understand the flaws and limitations of their tests, and if deemed important enough, might run a second round based on previous findings. In summary, when starting a round of website optimization:
- First, use data to uncover weak points in your users’ path
- Second, prioritize these weak points based impact vs. ease
- Third, develop key metrics such as hypothesis and KPI(s) for each test
- Fourth and lastly, implement the tests, uncover findings, test again if necessary. Repeat the process until you have zero problems (which will be never).