【Introduction】
Website analysis is still new, so our understanding of it may be biased in various ways. This article summarizes various common misunderstandings about website analysis that I found in my work. This is part two, and this part goes into some more detailed territory. For the first part, please see: The top ten misunderstandings and alternatives of website analysis (1). For the second part, please see: The top ten misunderstandings and alternatives of website analysis (2).
【text】
When I wrote the last article in this series, it was still on May 1st, and now it is August 1st. Time flies so fast, which is emotional.
In fact, in the first two episodes, the top ten misunderstandings have been discussed, and today we can only talk about the alternate ones. The reason why they are called alternates is that they are all very controversial areas. I am a member of the same family and I still only dare to make you laugh. But there are no substitutes for knowledge, and I hope to spark discussion, even debate, in order to gain real insight and truth.
Alternate Myth 1: There are standard benchmarks for website analysis
This is a place where there is a common misunderstanding. We often discuss bounce rate and time on site, so many friends will ask:
My website bounce rate is 60%, is that good? Or, the average time on site is 5 minutes, okay?
These are actually questions that I cannot answer, because website analysis does not have a so-called standard benchmark to refer to for these key metrics. All I can say is that the bounce rate of 60% is not the worst or the best I have ever seen. The same is true for the 5-minute time on site. However, as to whether it is good or not, these isolated data alone cannot answer the question. of.
The reason why there is no standard benchmark for website analysis is that the difference between websites is too great. First of all, the audience/traffic sources of the website are different; secondly, the functions of the website are different; thirdly, the content of the website design is also different; finally, the newness of the website is also different...
Therefore, there is no standard benchmark for website analysis! For example, we cannot say that a bounce rate below 60% is good, and a bounce rate above 60% is bad.
Now, you'd ask a better question:
If they are in the same industry segment or have websites with very overlapping audiences, can basic indicators such as bounce rate, time on site, PV/V, visitor loyalty, etc. be compared with each other? For example, Sina and Sohu, Tudou and Ku6, JD.com and Newegg, can they compare these indicators with each other?
I think they can be compared with each other. However, don't think that the indicator value of your website is worse than others, just because your website is not good. If Sina's bounce rate is 10% and Sohu's is 15%, will Brother Yang go crazy? No need, this does not necessarily mean that Sohu is worse than Sina. For that reason, the pages of Sina and Sohu are actually very different. Although they are both portals and they compete hard, they are still very different.
Similarly, the websites of Nike and Adidas, and the websites of Intel and AMD, they are all in the same tier (category), but they are actually very different. The size of these indicators cannot simply mean that one website is better or worse than another website.
Therefore, I have always insisted: Even for websites of the same category, simple numerical indicators cannot explain the quality of the website.
Then, you will ask again:
Since comparison cannot indicate good or bad, what is the point of comparison? !
Yes, of course! If you know your competitor's numerical situation, you can analyze it; you know that your numerical value is not as good as it and you can understand yourself. It is said that you can know the gains and losses by learning from people, and the same is true between websites.
Best, please don't create another misunderstanding, that is, since there is no standard benchmark, no matter what my numerical value is, it does not mean whether my website is good or bad, and I can sit back and relax.
I believe no friend would think so.
If your values are too outrageous and outside the normal range, that can still be telling. For example, if your website’s overall bounce rate is higher than 80% or even 90%, you should still pay attention. Website analytics love these anomalies.
Here are some extreme values from my experience (please note that these values are only valid for analysis using Google Analytics, other WA tools may have significantly different values due to different definitions and monitoring methods). If these values are exceeded, it may indicate that the website A more serious problem has occurred (but not definitely!)
Finally, I would like to remind everyone again that since each website is unique and the indicators themselves cannot be interpreted in isolation, there is no standard benchmark for website analysis.
Alternate Myth 3: Analyzing individual behavior is of great significance
I've seen some tools that record every visitor's mouse trajectory on the page. Each of these tools has its own strengths and weaknesses, but they are all powerful. Typically, these tools are intended for UED (UCD) designers, but do they have significant implications for website analysis?
Website analysis generally uses all (that is, no sampling at all) or large sample size data to analyze some behavioral patterns that website visitors converge on, and optimize the access experience of the most important visitor groups accordingly. Website analytics are rarely performed by studying individual visiting behavior. At this point, website analysis and website usability analysis are quite different.
If you have read "Don't Make Me Think", you know that after the website is completed, ask some ordinary people who have never used your website to complete some network access tasks you specified in front of you, and record them Determining their visit behavior is a very important method for testing and improving website usability. However, website analysis rarely adopts this method, which is to analyze and optimize through the data left by a visitor on the website. website.
The reason is simple, because the access situation of a large number of visitors is consistent with the normal distribution. It is possible that the access data of some visitors are distributed in extreme areas. If these data are used for analysis, the deviation will be large. For example, visitor A stayed on the website for up to 1 hour and visited as many as 100 pages. This does not mean that all visitors are like this, and analyzing a single visitor can easily lead to danger. Maybe you will say that I can analyze the behavior of a few more visitors, which will be more reliable. However, the problem is that compared with the number of millions of visitors, the number of individuals you can analyze is always limited, and the more individuals you analyze, the harder it will be for you.
Therefore, in my actual work, I will rarely use some very specific mouse trajectory monitoring tools, but I hope to have a mouse trajectory recording tool that records all mouse behaviors and uses different colors to represent the density of mouse behaviors. , which will be very useful to us and will be more valuable than the heat map we currently do. But it seems that no such tool currently exists.
Alternate Myth 4: Optimization solutions are the inevitable result of analysis
Website analysis focuses on analysis, yes, but analysis is not all of website analysis. The main purpose of analysis is to find problems, but analysis itself is not enough to help solve problems, or it can only solve part of the problems.
For example, when studying conversions, I often find that it's obvious that a certain page is losing a large number of visitors, but why is this page performing so poorly? Sometimes, based on experience, we can immediately think of the reason and suggest improvements accordingly; but sometimes, we don't actually know why the page is so bad. Even if we can think of a cause based on experience, this is not necessarily the real (or fundamental) cause.
Therefore, sometimes (in fact, most of the time is more accurate), the truly reliable optimization solution cannot be obtained directly from the analysis, but comes from the testing after you make suggestions through analysis. The advice itself is subjective, but the results after testing are objective (as long as you use scientific methods and processes). The cycle of website analysis does not end with analysis, but with testing, and testing is the way to go.
Therefore, although the picture below is a cliché, it is a truly important methodology.
Well, this article is finally over, and this series is finally over (there may be revisions and additions later). Thanks to all my friends for your continued encouragement! I hope everyone has different opinions, welcome to discuss, and welcome to argue!
Article source: http://www.chinawebanalytics.cn/top10-misunderstanding-for-web-analytics-part3/
Author: Song Xing