Super complete! A collection of information flow data analysis methods. With it, do you still have to worry about conversion?

Data analysis is difficult, as difficult as climbing to the sky.

I have written a lot about data analysis, but I have never written systematically about "data analysis methods".

So, I will write it systematically today, and you can choose the most suitable method according to your own situation.

It means to judge the delivery direction or delivery effect by calculating the proportion of a certain dimension in the total number of dimensions.

Formula: Proportion = value of a certain dimension / total amount X 100%

For example.

As shown in the figure below, this is a set of marketing data. Through calculation, we can clearly understand each region:

• How much did each region spend?
• How many conversions are there in each region?

Taking Beijing as an example, its conversion rate is lower than consumption, which means that the overall conversion rate is not good. Then we need to think about:

• What went wrong if the conversion was not good?
• Does the current consumption ratio in Beijing fit in with my current promotion strategy ?

In Hebei, conversion is higher than consumption, which proves that the conversion rate in this area is very good. Then we need to think about:

• Does the area need to add money?

By analyzing the proportion of each indicator, we can clearly understand the situation in each region.

This is the specific gravity analysis method.

Note: This method is more suitable for accounts with multi-product and multi-region promotions .

The backward method is a commonly used method in bidding promotion, but it is more often used in the formulation of strategic goals.

That is: based on historical data, the process of inferring transactions - clues - conversations - clicks - displays is reversed.

For example: the target number of leads this month is 50, and the reference lead rate is 50%, so we need 100 conversations to complete it; if there are 100 conversations, then by reverse calculation, 4,000 clicks are needed to complete the conversation; if the click rate is 5%, then we need at least 200,000 impressions to achieve the set goal.

Niu Gensheng of Mengniu once said: Only change the means, not the goals . By segmenting each dimension, when the goal cannot be achieved, we can clearly know which part should be optimized.

Keywords are the foundation of bidding promotion, so we can systematically classify keywords and carry out targeted optimization.

Generally, they are divided into the following four categories:

Most of these words are brand words, and they are high-quality words. For the better words, you can increase the volume.

For example: adding words, raising prices, adding matches, etc.

Words like this are mainly concentrated on product words and industry words.

A high click cost often indicates that the click traffic is large and complex. In this case, it is recommended to increase the volume conditionally, that is, to control the quality of the traffic while obtaining it.

The main operations are:

• Add words,
• Optimize account structure (make account flow structure more accurate)
• Optimize creativity (use creativity to filter out some impurity traffic)

In this case, there is often no concentrated part of speech. Usually, the next step can be determined based on the following two points:

• Is the average price high or low?
• Is the flow rate large or small?

If the traffic is large and the average price is low, it is often done by optimizing the page; if the average price is very high and the traffic is average, the price is lowered; if it is because the traffic intention is low, it is recommended to perform a matching operation.

In cases like this, most of the time, “a conversation was generated with just one click”, so we think that these are high-quality words and then increase the volume of operations, but it may also be an accident.

Therefore, in situations like the one mentioned above, we should increase the volume conservatively and only make large-scale volume increases after the situation becomes clear.

It means listing and collecting the core indicator data [ consumption, display, click, arrival, conversation, lead, transaction ] in the account every day, and then calculating some auxiliary data based on the core data, such as [ click-through rate, conversation rate, click cost ], etc., and discovering the type of disease by comparing the data of different periods.

For example, the picture below is a marketing process chart. Among them, the daily data changes are clear at a glance, but if you want to analyze it, you will find it too complicated.

Therefore, the data can be divided into two periods according to the date, and the two periods can be compared and analyzed.

Look at the picture below, which is a screenshot after data comparison. We can easily see the data changes between 6.16-6.22 and 6.9-6.15 and find the problem.

The pointer analyzes data between different dimensions to determine the optimization direction.

A single dimension can be mainly divided into: product dimension, time period dimension, device dimension, region dimension, and keyword dimension.

Take the time period dimension as an example.

We can analyze the time period data in a certain plan to determine which time period has better conversion and which time period has worse conversion; then modify the promotion direction based on the data.

As shown in the figure below, we can see that the conversion rates are high at 15 o'clock, 14 o'clock, and 19 o'clock, so we can increase our investment during this period.

Many times, the reason why data analysis is difficult is that we simply don’t know how to do it.

Understanding the above-mentioned data analysis methods can help us greatly improve the efficiency of data analysis and identify optimization directions.

Author: Little Rabbit , authorized to be released by Qinggua Media .

Source: Houchang College (ID: semlianmeng )