Time factors that must be considered in online marketing
Source: Shangpin China |
Type: website encyclopedia |
Date: March 1, 2012
Beijing website construction Shangpin China: Website production In data analysis, time is one of the most common and indispensable dimensions. In most cases, it is used to limit the scope and granularity of indicator statistics. At the same time, time factors will also affect some statistical rules and details of indicators. In some data analysis, we can easily ignore the impact of time factors, which may mislead the final conclusion.
It is found that in a data extraction requirement, websites will publish a lot of new content every day, and these new content needs to be recommended, otherwise it will be buried, so many websites will have modules such as "latest recommendation", and this data requirement is to analyze which new content should be recommended? The quality of new content on the website is uneven, and there is less data accumulation. However, the recommendation module needs to place new content with potential so that it can grow into popular content after full exploitation of potential. Therefore, data analysis is to find new content with potential.
If it is TOP 10 Seo website optimization The simplest way to recommend a list is to rank the top ten according to the number of visits or conversion rate of new content, but there are many things worth noting. For the conversion rate, you can refer to the article "Secrets behind key indicators". Here we mainly discuss how to rank the list according to the number of visits to content, and how to select the summary data of the past week, What should we pay attention to? As you may have thought, the reason why new content is exemplified here is that there is a publishing time for new content, just like a person's birth date. The time interval from the publishing time to the current time is the duration of the content, which can also be regarded as the life time of the content, just like a person's age. The longer the duration of the content, the more data will be accumulated, and the greater the chance of obtaining high traffic. If we compare the total traffic of content published at different times in a week in that week, those will fall into the trap of dislocation comparison, or "Mismatch".
A vivid analogy is the duel between a newly enlisted recruit and a veteran. Although the recruit does not have no chance to win, perhaps the recruit is born brave, or has a new momentum of fearing the tiger, and can defeat the experienced veteran at one stroke. But in most cases, this is less likely. It is an unfair duel, In data analysis, we need to try our best to avoid such unfair duels (comparisons).
Content and commodity analysis
We need to find some ways to avoid the impact of this time factor on the analysis results. Usually, when we select comparison objects, we need to control that all comparison objects have the same duration. For example, we compare the popularity of new content, uniformly select the data of nearly a week, and discard the previous data for the content released earlier, However, the content just released in the past week will not participate in this comparison until the complete data of the week is available. Although this can ensure that the comparison is on the same baseline, it undoubtedly delays the conclusion of the evaluation. For some content that has been eye-catching at first, it cannot be found in time. Therefore, the method of statistical unit time indicator performance is adopted here, that is, the duration of each content is calculated according to the release time of the content (usually accurate to days), Then divide the total access volume of the content by the duration to get the content access volume per unit time, and then compare it.
In fact, it may be common for such errors to occur in daily life. When I went to Google Analytics to look at the data a few days after my new blog post, I found that the new post page was in a relatively backward position, not because no one really looked at it, but because the GA showed the data of the last month by default, The new content of the report sorted according to Pageviews cannot be rushed to the top in a short time. For those websites with new content or products that are not frequently launched, operators may be more clear about what is new content, so it is not easy to fall into a trap when analyzing through some artificial identification and adjustment. However, for websites with hundreds of new content every week, the occurrence of such errors is likely to bury some high-quality new products:
The above table takes the access data of 5 newly released content for nearly 10 days, and adds the duration days of the content since its release. We calculate the average daily access by dividing the total access by the duration days, and then rank them in descending order by the total access and the average daily access respectively to get a completely different ranking. If we rank 1, we may completely ignore the strong performance of D content, and the ranking after weighing the time factor allows us to more accurately grasp the potential new content.
The above methods are also applicable to the analysis of commodities on e-commerce websites. Many e-commerce websites hope to select commodities with sufficient potential for key marketing among new commodities to create so-called "blockbusters", so as to further promote the growth of orders and increase sales and profits. The selection of potential new products requires a keen sense of smell and vision on the one hand, and data analysis on the other. At this time, we have to consider the impact of the time factor mentioned above. Remember that 20 items sold in a month are not necessarily inferior to 50 items sold. The key is when these items are put on the shelves, Only by using effective methods can we find commodities with real potential and valuable growth points.
You should know that any website content or product is not enduring, but has its own life cycle. Therefore, smart website operation is always looking for new growth points. If time factor is not taken into account in data analysis, potential products and content may be suppressed by "honed" product content for a long time, As a result, the metabolism of websites is too slow, and then lags behind other websites.
User analysis
When conducting user analysis, we also need to pay attention to time factors, such as user RFM analysis, user loyalty value scoring, user life cycle value, etc. These models based on the analysis of users' persistent behavior over a period of time are easy to fall into the trap of time. We can't expect a new user who has only registered for one week to visit more frequently than an old user in the past month, because you only gave him 7 days of time, and he was challenged by a user who has sufficient 30 days of time; Similarly, you should not compare the consumption times and amount of a new user who has only used for one month with an old user who has been using for a long time in three months or six months, because they are not on the same running line. However, new users have potential, which means that they will grow into more valuable loyal users. Therefore, we need to eliminate the influence of this factor in the marketing for users. We also use the method of dividing by the duration of users using the website (calculated from the time of users' first visit or registration) to calculate the indicator performance of unit time, Use the RFM model to see the difference in user evaluation before and after considering the time factor:
As shown in the above table, if the RFM model selects data of nearly 100 days to analyze users, the statistics of "duration" is also added here, that is, the number of days from the registration of users to the current one. If the registration time of users is 100 days ago, the duration of users in the statistical cycle is 100 days (the maximum period). The most recent purchase interval (R) in the three indicators of RFM is not affected by the user duration, so there is no need to change when considering the time factor, but the purchase frequency (F) and consumption amount (M) are affected by the duration. You need to divide the duration to calculate the unit time (in this case, days), that is, whether to consider the time factor for each user in the table Index transformation before and after. From the comparison before and after the transformation, user 1 has obvious advantages in purchase frequency and consumption amount before considering the time factor, because it is an old user with continuous use. However, after the data transformation, user 2 shows higher stickiness and value, that is, although user 2 has not used the website for a long time, it is better than user 1 in purchase consumption per unit time, Let's take a closer look at the effect before and after considering the time factor through the radar chart:
After the standardized scoring of the data in the figure, the blue line represents user 1, the red line represents user 2, and the dotted line indicates that the time factor is not considered. The implementation represents that the time factor is considered. It can be seen that the value of user 2 is significantly magnified after considering the time factor. From the figure, it can be seen that the expected value of user 2 is better than that of user 1. If we do not consider the impact of time factors, the analysis results will produce obvious deviation, which may mislead the correct evaluation of users.
In fact, the time factor mentioned here is still a matter of following the principle of comparison. The objects of comparison must be comparable, or the results of comparison will be meaningless.
I haven't updated my blog for a long time, because I don't have time to think about and sort out some new content due to the changes in this period of time. The time factors that need to be considered in the analysis mentioned in this article have actually been encountered in many cases, especially when performing a breakdown analysis on the statistical indicators summarized within a time period, it is necessary to pay special attention to whether the time periods of each breakdown item are consistent, hoping to enlighten and help everyone.
This article was published on Beijing website production Company Shangpin China //ihucc.com/
Source Statement: This article is original or edited by Shangpin China's editors. If it needs to be reproduced, please indicate that it is from Shangpin China. The above contents (including pictures and words) are from the Internet. If there is any infringement, please contact us in time (010-60259772).
