How can product managers be insensitive to data?

Product managers must be sensitive to data, from the data to read out the trend of the product, the reasons for the change and how to deal with. Web site Traffic Change analysis is the basis of the site analysis, I will simply say my understanding of the changes in the volume of visits.

Clear indicator definition

Before analyzing, we must first understand the detailed definition of each important index. For example, the access may be in session, the independent visitors may be 24 hours the same IP only calculate once. Sometimes different statistical plug-ins define these metrics differently, so figuring out a definition allows us to have a clearer idea of what the metrics represent. The amount of traffic here can be easily understood as PV, an HTML request from a browser.

Contrast rather than take for granted

Figure 1 Blog Traffic changes

Figure 1 is a screenshot of my personal blog's recent traffic changes. What happened on the day of the highest point in the picture, and why did the traffic suddenly increase a lot? The answer is the day before I published a blog, the second article was two science and technology blog to use, due to the submission of the original link, so the guide back some traffic, you correct?

I think a lot of people must have guessed that I published a blog, but few people would think that the increase in traffic is due to the science and technology blog outside the chain. Why simply think that blogging is causing a rise in traffic? Because on the one hand it is common sense, on the other hand there is no more data to compare.

When we make judgments, it is easy to rely on our own common sense to make assumptions, which are accurate and efficient in most cases. But we cannot rely entirely on common sense, and the analysis of data changes needs to be contrasted. By contrast, if you find that there was no significant increase in PV on the same day that I was posting, you would wonder if there were any other reasons for this PV to rise abruptly. It could be that the blog was recommended, or it could have been a blog event that generated a lot of people visiting me via search engines, so you might be closer to the truth. So in the face of such problems, try to find more data and information to help judge the analysis, seemingly simple changes may be a trap, contrast is very important.

The above summary is: The data analysis before the definition of indicators have a clear understanding of the analysis of changes to the previous data or competitor data, and then make a conclusion.

Comprehensive analysis to obtain the optimal opinion

Conclusion should be able to guide our action, the purpose of website analysis is to improve the site, just analyze what is, why not much meaning. For my blog, the goal is more people can see my article, then the simple data given me the guiding significance is to consider to improve the frequency of the dispatch, and focus on technology blog recommended this channel. Different sites have different purposes, the Electronic business site hopes to improve conversion rate, increase sales, news sites want to increase the number of visits and comments, improve advertising revenue, to the data changes and the fundamental purpose of the site, such guidance is the most effective.

In order to be simple, the above is only the traffic, there are many indicators such as the number of independent users, the average user stay time, the average user access depth, different page bounce rate, and so on. Only relying on the number of visitors can not draw accurate conclusions, the indicators should be linked to the analysis. For example, you find that PV is very high, but the jump rate is also high, indicating that you can promote doing well, do attract users to the site, but landing page did poorly or inconsistent with user expectations. In that case, you should consider making improvements to the landing page or making the promotion more accurate.

A question of a face

Finally ask you a question: a social networking site in mid-July, the number of visits have dropped significantly, why? Can you think of several possibilities?

I think several possibilities:

1. Does the volume of visits refer to changes in the statistical tools of the pv?7 month? Is there a change in the definition of the access quantity?

2. If possible, is there a decline in the mid-July of the previous two years, if so, compared to the previous data? If it declines, it may be a change in the user base as a whole;

3. If possible, subdivide the data, by region, by page, by user, you can get more information.

4. To determine the mainstream user base of the social networking site, if it is a student, then July summer vacation home, may result in a decline in the number of visits.

5. Was there any adverse reaction to the release of the new version of its website in mid-July?

6. Has the competitor released a new version that attracts users? (such as on the search engine to buy our site's main keywords)

7. Has the media previously reported bad news about the website (such as the website stealing user's privacy) affected the user's behavior?

8. Was there a social event that led to a decline in the number of visits in mid-July? (for example, after the Wenchuan earthquake 3 days)

Can you think of any other possibilities?

Summary

Analysis of the changes in traffic to pay attention to the definition of indicators, subdivision, contrast, in the face of change to consider their own, competitors, user groups and macro environment, combined with different indicators for comprehensive analysis, at the same time the optimization of the site to give guidance, after all, this is the ultimate goal of analysis.

Originally wanted to write access to change and the source subdivision, did not expect light traffic changes on the said so much, that access to the source of subdivision on the next article to say well ~