Engagement is a special metric. Avinash believes that Engagement should not be used as a metric for website analysis. It is more like an excuse for not seeking real indicators to measure the success of a website. He listed several shortcomings of Engagement:
Engagement is different for different websites. It is not a unified measurement and it is difficult to define it with a unified standard;
Engagement is defined more from a qualitative perspective. Many user participation behaviors cannot be quantified well, so it is difficult to accurately measure;
There is no criterion for judging how much user engagement is good for a website;
Because there is no standard definition and there is uncertainty, it cannot be used as a reference indicator to measure the overall performance of the website.
Therefore, we can think of Engagement as a non-standard metric. But after avoiding the misunderstandings mentioned by Avinash, maybe Engagement is not useless. At least it can be used as an analysis indicator of the website itself to measure user activity.
How to reasonably define Engagement
If you want to use Engagement to measure website user activity and use it as a long-term indicator for trend analysis, then you must first determine when defining that the behaviors classified as Engagement are relatively fixed and will not occur in the short term. change. For example, for a forum, the behaviors of "posting" and "posting" are continuous interactive behaviors, which can be used as criteria to identify whether a user participates; while an active button or promotion link on a website cannot be used to identify engagement. Standard, because promotion activities generally have a time limit, and the user's Engagement will decrease accordingly after going offline, which will lead to instability of Engagement and lose the meaning of analysis (or you just want to analyze the promotion separately. User participation in activities is another matter).
Here are two examples to show how e-commerce websites define Engagement:
1. User’s registration behavior;
2. User’s shopping cart or purchase behavior;
3. Users’ pre-sales and after-sales feedback behaviors;
4. Users’ evaluation of the product.
Let’s take a look at some behaviors that can be defined as Engagement in Weibo:
1. User registration behavior;
2. Publish a new Weibo or forward a Weibo;
3. Post comments on Weibo;
4. Pay attention to new users.
It can be seen that the definition of Engagement of e-commerce websites is very different from that of Weibo. This is why Engagement is a non-standardized measurement. The pros and cons of Engagement cannot be compared between websites. Therefore, Engagement should be used as an internal analysis indicator of the website and is only used to measure the changing trend of user activity on the website itself.
Even a website's Engagement has different definition methods. For example, I can also add the action of "favorite Weibo" to the definition of Engagement on Weibo. So my suggestion is that for the definition of Engagement, a website should form it at the beginning. As a standard, if the nature of the website has not changed significantly, do not modify the definition easily. Otherwise, even if it is used as an evaluation indicator of the website itself, it will not work well.
It should be noted that Engagement is just a series of user actions or behaviors (Actions), not the output results (Outcomes) of the website. It can only be used to measure the frequency of website user activities, and then analyze the recent activity level of website users, but cannot be used to measure the effectiveness of the website.
Analysis based on Engagement Index
Here I first recommend an article by Eric T. Peterson on Web Analytics Demystified - How do you calculate engagement? , which provides a very comprehensive and exciting analysis of Engagement.
Analysis is best combined with examples, so here is my blog as an example. If I define my blog’s Engagement to include the following behaviors:
1. Users post comments;
2. Collect or share my articles;
3. Subscribe to my blog or follow me;
4. Time on site exceeds 3 minutes.
First of all, it involves the issue of data acquisition. In fact, through click stream + event tracking, most user behaviors can still be tracked, so we will not go into details here.