Three ways to measure the effect of network marketing on the applicability of indicators

Intermediary transaction SEO diagnosis Taobao guest Cloud host technology Hall

In the previous series of "core metrics for Network marketing effectiveness", we introduced what Engagement,engagement included, how to define them, and how to monitor them in a technical way. This issue will be further explored to explore the applicability of indicators.

Because of the flexibility of indicators, it has very high research value in at least three fields. First, it helps to measure the overall (macro) effect; second, in the knowledge of the quality of the flow, it can measure the impact of the site on users (that is, the quality of the site we call it), and its third, in the knowledge of the quality of the site, it can measure the quality of the site traffic.

These three questions are almost the core area of our marketing analysis. Let's take a look at the first question: How to measure the overall (macro) effect through the indicators index?

Indicators three ways to measure the overall effect

The indicators measure of macro-effects is based on the assumption that a large number of users appear to be chaotic behaviors that actually reflect the extent to which the site affects the user (attraction). The greater the amount and degree of user behavior, the more strongly the Web site affects users. That's not hard to understand. The same two news stations, a station average per user visit to see 10 news, B station only 5, it is obvious that a station to attract more users. Indicators is essentially used to describe the chaotic behavior of various users, so the number and intensity of indicators is equivalent to the ability of the site to affect users.

So how do you measure the amount and intensity of indicators?

There are usually three simple ways to measure macro indicators. The first approach is to observe global indicators metrics that can be either standard or custom. The second method is to sum up the individual indicators index according to the important degree of customization, and to form indicators index to measure. The third method is to define the importance of each index according to the mathematical relationship between user behavior and marketing target, then the weighted sum into indicators index. As long as the user's behavior is monitored properly, three methods will not be difficult to operate.

Method One: Analysis of global Indicators index

The first approach, which observes the overall indicators metric, is the most common observation of bounce Rate, pv/v, or time on Site. About Bounce Rate, we have the most problems. In almost all classes, when I explain Bounce rate related knowledge, there will be students ask: Bounce rate How high is good? I can simply reply, if your Bounce rate no more than 65%, then it is really worth celebrating. But this answer is not what I like, if Jingdong bounce rate is 65%, then they will be suffering Tong. However, if in the marketing campaign, the site's bounce rate is really above 80%, then we should draw what kind of conclusion?

Conclusion A:

Damn, this is a total failure of the activity, traffic sucks, the site is also pulpy.

Conclusion B:

The website is successful, but the traffic has a huge problem.

Conclusion C:

Traffic is sure no problem, but the site is bad.

Which of the three conclusions would be correct? The answer is, it's possible. However, in specific cases there must be only one possibility. How do we judge bounce rate?

My method is very simple--according to the source of traffic breakdown bounce Rate, and then determine whether the flow out of the problem, or the site out of the problem. To solve this problem I will introduce the concept of reference frame, that is, flow can be used as a reference standard flow source. We all know that Web site traffic sources are diverse, where the quality of the best flow is the natural search flow (Organic). Perhaps you will be disappointed to ask why it is not direct (directly flow), because direct is not really directly traffic, but for this web analytics tool all the traffic is not able to judge the origin, so direct may mix a lot of traffic, the quality of these traffic is almost certainly not as natural search traffic.

Why is organic the best quality? In fact very simple, organic traffic is not cheating traffic, and reflects the real user access needs. If a website does not have what organic traffic, then the traffic quality second good flow is the paid search engine traffic. These two kinds of traffic can be used as reference for us to measure bounce rate. Now, let's look at a few of the following, and we'll draw a separate conclusion:

Situation A:

Organic flow of bounce rate good, only about 40%, but the overall site bounce rate is 80%.

Situation B:

Organic flow of bounce rate and site overall bounce rate almost as high as 80%.

Believe that these two situations, the reader has a few minds. A situation, that the site for those who are really interested in the attractiveness of the user is good, 40% bounce Rate Standard, but the site still has up to 80% bounce Rate, indicating that other traffic is probably not the people you want to get. b, if even organic flow has a very unsatisfactory bounce Rate, then that the site really do almost mean. Flow reference system gives us a very clear judgement is the flow or site problems, but the use of this method has a prerequisite, that is, organic flow can not be too little, if organic traffic is very small, can do the replacement of the flow is paid Search. But if there is no paid search traffic, find the sources of traffic that you think are reliable. However, regardless of the performance of the reference frame flow, a marketing campaign, if the overall bounce rate higher than 80%, it is certainly not a particularly ideal situation, either traffic or site problems, or there is a waste of marketing costs.

Pv/v and time on site are more difficult to benchmark than bounce rate. However, as far as I can see, it is almost certain that the higher the pv/v and time on site of a website, the more the marketing audience is affected by the site. I used to think that there were some sites pv/v or time on site too big to be bad, such as government service sites, too high pv/v or time on site to show that people might not find what they were looking for. But after tracking a lot of websites, I found that people's patience is really very limited, if a site is not easy to find the content they need, even important sites, they will turn to Baidu, rather than in the experience of the bad site crazy pursuit. The average pv/v is greater than 3, or the average time on site is greater than 2 minutes to show that the site itself has a minimum of appeal to the user.

When looking at the global, custom indicators metrics are more common here. For example, the number of clicks on a promotional item (or the number of times to open a promotional item page) is viewed as a global indicators in a promotional event, and the actual amount achieved is compared to the pre-set target to see if the marketing effect is achieved. This method is very quick, but also very extensive. If bounce rate is above 90%, but luckily, the number of clicks on promotional items is more than twice times the pre-set target, is this marketing campaign good or bad? A single view of indicators, always encounter some uncomfortable trade-offs, and always difficult to really take care of the overall. Then we have to consider other new methods.

Method Two: Weighted summary of single indicators index

The second method, the single indicators index in accordance with the importance of the definition of a weighted sum up to make up for the first method of partial deficiencies, after all, the macro-indicators is a specific one of the indicators index synthesis results. This method has three steps: First, list all the indicators specific indicators, and then according to the importance of each indicator in your mind, assign a weight to each indicator, and finally multiply each indicator by the weight, add total. Add total value, that is, indicators Index.

Different types of marketing choice of specific indicators and the weight of the arrangement is certainly not the same.

The second method is more comprehensive than the first method, but the allocation of weights is very subjective, and some commonly used global indicators, such as bounce Rate, pv/v or time on Site, can not be included, and have to become two sets of parallel measures. The latter is not a big problem, but the challenge of weight distribution seems easy to be the boss and the client. However, in fact, this approach is currently the most used method, because it is relatively simple, and for an advertiser, fixed some common behavior of the weight value has many advantages, it is very clear that the marketing campaign should be the direction of effort, It also helps to campaign horizontal comparisons between different stages or similar campaign.

Method III: Weighted summary of the indicators index under the transformation relationship

The third approach, which makes some improvements to the second approach, is based on the idea that in a marketing campaign, the behavior of the user in a seemingly chaotic manner actually has a fairly definite proportional relationship to the ultimate goal of ultimate achievement (e.g. transformation). Although these ratios do not have the same value for a variety of marketing activities and sites, these ratios are relatively stable for a defined activity or website.

For many e-commerce sites, for example, unless there is a major change in the category structure, their "shopping cart to the actual purchase of the conversion probability" is more stable, such as on the benchmark of 40% floating around. Brands that do not need to sell goods to promote marketing, in fact, also applies to this relationship, because almost all of these marketing to explicitly want users to take action (such as applying for trial, sharing to friends, etc.) as the ultimate goal, these specific actions and E-commerce site purchase behavior is actually not the essence of the difference. Now, let's assume that an ecommerce web site has the following transformation rules: every 100 new registered users will generate 4 orders, each 100 IPV will produce 18 orders, and every 100 times put the goods into the shopping cart will produce 20 orders ... We can get the following table (table 1).

In order to compute the relationship in the following, we have each indicators index into 1 as the benchmark, we can get a series of proportional relations, with this correspondence, we can easily define the indicators weight of all important events before the transformation occurs. If we convert 1 to 100 points, it is quite easy to calculate the indicators index of each index according to the actual value of each different behavior (table 2).

The third approach looks more scientific than the second, and Google Analytics's page value setting is similar to this one. But this method can not be said to be the reality of the full true feedback, it still has some problems. Since this method is based on the assumption that all actions have a direct contribution to the final transformation, but transformation must be a process in which there is a relationship between the different behaviors of the user at different stages of the transformation (positive promotion or negative interference), which is not covered by this method.

Interestingly, for this approach, you will find that, because the internal transformation of the site is also step-by-step, different processes on the final transformation of the value is not necessarily the same, so it seems to be fully applicable to attribution modeling way, using different modeling (such as linear, Descending or Middle high both sides low) way, to the user different behavior weight value is also not the same, you can also use attribution modeling model of the idea according to the actual situation to the different indicators indicators to empower, so that you can be more close to the business reality you expect. But the difficulty of operation looks rather gloomy.

Although three methods are not 100% indicators to the true extent of the user's authenticity (but the full reappearance of reality is only an ideal state), it is helpful for the quantified indicators we expect. For a marketing campaign, using these methods can tell us from the perspective of the process whether it is moving along the route we expect.

Author: Sing via: Titanium media