Search Promotion | Four Principles of Data Analysis

Many friends want to learn data analysis, so I will write down my own data analysis ideas for reference only.

Data analysis is actually a very empty concept. The word "analysis" is too broad. From a theoretical perspective, I think data analysis must first have some principles. That is, what should you analyze and what should you focus on. Tell us what you think about this.

1. Pareto Principle

First, one of the principles of data analysis is the 80/20 rule.

To put it bluntly, we have a lot of words in our account, but only 20% of these words are displayed, and the remaining 80% are not displayed. We should manage these 20% of words first, and then look at the words that are not displayed. This is the relationship between primary and secondary. And if those 80% of words are not displayed, there must be a reason.

If after adjustment, it still doesn’t show up, then it’s the word itself. At this point, these adjusted words are still not displayed, so we can actually ignore them. Once it is in the account, there is no need to delete it.

2. The Four Quadrants Rule

Many people know this, the so-called four-quadrant rule, which means that there are four types of words in our account. High consumption and high conversion, high consumption and low conversion, low consumption and low conversion, low consumption and high conversion.

(Photo source: Aichi SEM)

The core of data analysis is to grasp the control of words, so the 20% of words displayed in the account should be paid attention to regularly. The core of the focus is the consumption and conversion of words. Let’s look at them one by one.

High consumption and high turnover. For example, the word for medical expenses is a conversion of high consumption and high expenditure. This type of words accounts for 80% of the account's consumption, but the conversion rate is also relatively good. For this type of words, we need to: pay attention to them first; pay special attention to their matching; calculate the conversion cost of the words in a week or ten days; strictly control the inflated prices; we can set prices for these words in a gradient manner, with precise high prices, short phrase medium prices, and intelligent core low prices...

In short, you should pay special attention to this type of words and see whether you can accept the final conversion cost. If you can, you don’t need to make any adjustments. If you can’t, you need to lower the price appropriately and just maintain the ranking of second, third, or fourth. Regarding high-consumption and high-conversion keywords, we also need to look at them based on the keyword conversion table. Its processing is also relatively flexible, which will be explained with examples in the following text.

High consumption and low rotation . For this type of words with high consumption and low conversion, the first thing to do is not to lower the price, but to look at the page and search terms. If there is no problem with the page and search terms, then lower the price. At the same time as the price is lowered, check whether there are any single words with a large number of clicks. If so, consider whether it is a malicious click.

For high-consumption and low-conversion words, if there is a problem with the page, adjust the page first. If there is no problem with the page, we usually lower the price first, and then adjust the matching mode. It is necessary to clarify the order of adjustments. Why lower the price first and then adjust the model? Because high-consumption words are usually high-intent words, the matches are generally narrow, and the search term problem will not be too big. So first reduce prices to control consumption, and then adjust the model.

Low consumption and low conversion. These words are actually so-called useless words. For useless words, first of all, they have low consumption, which means they have few clicks or low average price. The low average price indicates that the word type intention of this type of words is not strong, such as the medical word postoperative. Fewer clicks, excluding the factor of narrow matching, indicate that the search volume for this type of word is not very large, so fewer clicks and low conversion rates are not very relevant.

Therefore, for words such as low consumption and low conversion, we first check the page and search terms. Secondly, we can moderately relax the matching or add a little price to increase the number of clicks. Only in this way can we better weigh the effect. Sometimes, some low-intent words in an account can occasionally bring conversions. If there are no clicks or few clicks, it is acceptable to increase the price by 10% to 20%.

The low-consumption and high-transfer word can be said to be the best word in the account. For this type of words, the page should generally not be changed, but the negative words should keep up, such as negative words once every three days, and actively expand the words. To keep this type of keyword ranked in the top three, it is not recommended to adjust the price too frequently. Increase the display and clicks of such words as much as possible. Don’t make drastic changes to the matching, but make adjustments flexibly.

The above is the idea of ​​the four-quadrant rule. There are many explanations on the Internet, so I will briefly summarize it here. In short, search data analysis is to analyze the consumption and conversion of words. The 20% of words displayed in the account should be divided into these four categories, and then these four categories of words should be broken down one by one.

3. Analysis cycle

As mentioned before, a stable account does not really need to be adjusted, but if it is an unstable or new account, it can be adjusted frequently. As for the data analysis cycle, according to my personal habit, for large accounts, I will make a conversion table and analysis every three or four days, and then make adjustments. For a small account, do it once a week . Of course, some people may only do a comprehensive analysis every 10 days. There is no good or bad method.

When it comes to the cycle of data analysis, novices are too hasty while veterans are lax. In fact, the doctrine of the mean is the most appropriate. Taking myself as an example, I review a word once a week, make a keyword conversion table, and compile a data month-on-month table. Analyze and adjust on a weekly basis. The analysis methods of different channels are similar, but the characteristics of each channel are different. For example, 360 Match and Baidu are different, so the characteristics of each channel must be grasped and strategies must be formulated based on the channel.

4. Elements of Data Analysis

We always talk about data analysis, but what exactly is the object of analysis? What are analytical indicators?

For search, there are actually many objects to be analyzed. The most important one is keyword analysis. In addition to keywords, creativity, matching, and price must also be analyzed.

In addition, is it necessary to analyze units and plans? If your account structure is clear, then the units and plans must be analyzed. If your account structure is not clear, then the units and plans do not need to be analyzed. For example, in medical treatment, diseases are often divided according to plans, and analyzing plans is analyzing diseases.

Keyword analysis is often combined with keyword conversion tables, prices, and matches. Creative analysis is mainly based on a week, looking at the month-on-month display and click-through rates of the creatives, and then slightly optimizing the creative copy and images.

For plan and unit analysis, you can usually make disease conversion tables and part-of-speech conversion tables, but the prerequisite is that your account structure must be very clear, and a clear division must be achieved when segmenting the words, so that the table produced is accurate.

The above are some basic principles for search data analysis. In short, managing an account requires care and attention, grasping the key points, and paying attention to data changes. After you have been doing this for a while, you will become sensitive to the data and know how to make specific adjustments.

Author: Mubibai

Source: Mubibai Internet Marketing