The meaning of real-time data in web analytics

With the continuous development of the Internet, information more attention to real-time, micro-bo popular, search engines have launched real-time search function, but for the site analysis of real-time data is more meaningful?

Actually look at the data to look at the report people often want the data the more real-time the better, they want to master the site every hour or even every 10 minutes of change, able to the current situation of the site at hand, can find problems and quick response. But what if you ask them what can we do when they know the real-time changes in the site's data, or when the site hits a sudden surge or surge in a certain period of time? I guess most people can't answer. Just a while ago in the site to do real-time data statistics related work, so some ideas to share here.

Advantages and disadvantages of real-time statistics

Whether real-time statistics are useful or useless, first look at what needs to be done to get real-time statistics, and what real-time data can bring to us, which is the pros and Cons of real-time statistics.

First of all, from a technical point of view, it is clear that real-time data statistics need more resources to occupy, because the site analysis of most of the data need to be computed from the click Stream data, and there is no ready-made results data can be obtained straight line display. The data obtained from the click Stream needs to be calculated and aggregated, which is no doubt that these operations require more cost, especially for large web site large data processing, while real-time data increases the complexity of implementation, and may increase the data inaccuracy to some extent.

But real-time data statistics can show the ability to process data at the technical level, while providing richer report presentations, and even using dynamic trend charts for real time refreshes on the report, which naturally goes without saying, so sometimes many technicians are happy to do the work.

From the point of view of data application and analysis, at present, many real-time data statistics are used to show the real-time traffic changes in the site, which time period of the highest number of visits, or the site's overall activity is the highest, at the same time can analyze the daily traffic or the number of users distribution, but these analysis of the site in the end how much significance? What can we do even though I know the Web site has the most online users at 8 or 9 o ' night? The stress tests on the site obviously don't need to be done this way.

So personally think that real-time statistics more is the real-time status of the site monitoring, for analysis, not much practical significance, as to the site optimization and decision support how much effect, at least I have not thought.

Note to Avinash Kaushik in the book mentioned a sentence: "Real-time data:it ' s not really relevant, and It's expensive to Boot." I really agree with this statement. Many people will feel that getting real-time data will be more conducive to real-time response, fine-grained data also provides a more detailed base for data analysis, and we can do more analysis based on this, but we need to see how much the cost and true value of real-time data can be. Avinash Kaushik also enumerates the negative effects of 5 typical data acquisition in real-time, which can be summed up in the following points:

Do not blindly pursue the amount of data, more attention should be paid to the quality of the data, and through effective analysis to reflect the value of the data, not in line with the principle of 10/90, real-time data on the acquisition of the cost obviously can not be reached with the analysis of the value of 1:9; too much attention to real-time data will be binding on the choice of analysis The inability to use truly excellent web analytics tools, technical system resource usage, task scheduling, and complex processes may increase the inaccuracy of the data to some extent.

Of course, real-time data is not a bit of value, but for its cost considerations, there is no need for each analysis indicators in real-time statistics, or spend a lot of energy to focus on real-time data.

Value of real-time data

In fact, whether Google Analytics or Baidu statistics, have provided some indicators of the hourly statistics. Baidu Statistics will be real-time data statistics on the site shows that the user as long as a login can see the day of PV, UV and other data changes trend:

And in Google Analytics, we may find that GA is usually shown in the days of each metric, but in fact, GA also has the whole point of statistics, but the hidden more "deep" in the Visitors-visitor trending, in visits, Pageviews, Bounce rate and other reports will be found in the upper right time range Select the following time to summarize the granularity of more than one option--hour, you will see the trend of data changes every day after you choose:

Real-time data is not useless, avinash Kaushik that real-time data can create value when a company has fast analytics, fast decision-making, and fast execution capabilities. I'm here to give you a few of the applications that I think if we can get statistics per hour, then we can know in which time the site has the highest number of user visits, you can do some promotional activities in this time period, and through real-time data statistical analysis of the effect of activities to make a quick and reasonable response. For example, "second kill" activities need to be in a very short period of time to complete the statistics and show the results, of course, the prerequisite is the need to be in the background of the statistical system can withstand conditions.

Finally, borrow Avinash Kaushik a sentence as a summary: if only to see real-time data for real-time statistics, rather than based on real-time data to make corresponding action, then real-time data is quite expensive.

It's your turn, what do you think about real time data analysis? May allow me to realize the real time data generated in addition to real-time monitoring outside the more valuable results, welcome message comments.