6 analysis methods to teach you how to quickly diagnose SEM account performance
|
Only data can tell whether your promotion account is good or not. This article is for those who need to quickly understand an account. These six methods are: "one-nine" distribution, four quadrants of consumption and conversion , conversion decomposition , bubble chart analysis of core word interactivity and ROI, comparative analysis of core word interactivity and ROI of single-page promotion, and word correlation analysis between common words and brand words. The more detailed work of optimizing each keyword will not be involved. Method 1: "One-Nine" Distribution Law The long-tail model describes the distribution of keywords within a SEM account from the perspective of word frequency distribution. The SEM accounts in this world basically conform to the 28 or even 19 distribution rule. For example, the distribution of keywords in the following account: The top 250 words for this account (which account for only 6.8% of all traffic -generating words) brought in 88% of the traffic and 86% of the conversion revenue. Quite typical. Another account also reflects this pattern (as shown below). Among all the keywords that generate traffic, only a very small part of the top keywords generate significant traffic. Therefore, using this model can help us focus on keywords: Account optimization is actually very easy. If we focus on the top words that can bring the most traffic, we have already grasped the key points! Combine these words with the four-quadrant model that will be mentioned soon, and your optimization goals will become much clearer. Method 2: Four-quadrant model of consumption and conversion Everyone must have seen a model like this. This is the most commonly used model in Internet marketing analysis (in fact, all business analysis). Used to distinguish things in two dimensions. In the macro research scenario of SEM keywords, this is a very good model. The advantage of this model is that it is very easy to understand and is very helpful for grasping the macro level of keyword performance. The disadvantage is that this model is designed to help you grasp the macro picture, but in order to understand what is happening behind these four quadrants, you have to immediately go into the details of the micro words. For example, according to the four quadrants above, the following specific word breakdown is made: So, you see the details of many keywords, but you still don’t know how to optimize them. Because, judging from the meaning of each word itself, there are essential differences between different words. The meaning of some words determines that they are easier to convert (for example, brand words, where to buy, price, etc.), while other words are not easy to convert in themselves (such as general words). Therefore, according to the superficial requirements of the four-quadrant model, finding ways to reduce costs for those with high costs, or finding ways to increase conversions for those with low conversions, or expanding keyword sources for low-cost, high-conversion keywords, etc., are more of guiding ideas than optimization strategies, and it is difficult to help you form specific actions. So, as a reminder, you still need to go into the details of each word to consider why they are in different quadrants. Nevertheless, it is still very helpful for you to quickly understand which keywords are not good enough, especially the head words found in method 1 and those with core roots. Method 3: Transformation and decomposition Transformation decomposition is the easiest model to operate (although not easy to achieve accurate data), but it is also the most misleading model. We will follow a similar process of "exposure -> click -> traffic -> consultation -> effective consultation -> registration -> purchase" for each conversion step of the entire SEM account (or the account's plan, unit, etc.), so as to quickly locate where the problem occurred. It is precisely because of this method that we have discovered many times that the fundamental reason for the poor bidding effect is not the bidding itself, but the problems with the traffic in the landing page, details page, customer service consultation, etc. after the bidding. In this way, the hard work of blindly optimizing the account must be quickly transferred to the landing page or wording. This method is what everyone who does SEM analysis and optimization will definitely adopt. But from an operational perspective, there are two main problems to avoid (although they are not problems with the model itself). The first problem is that the conversion funnel model requires accurate data for the entire process, but it is difficult to obtain all of this data accurately. Especially for industries that require a certain conversion cycle, such as finance, training, and medical care, consumers have a fairly long decision-making cycle. Therefore, we can compare the conversion statistics in real time back to the level of "effective consultation". In order to solve this problem, our ROI calculation often starts from: Delivery cost -> Final revenue Transformed into: Delivery cost -> the latest conversion of the day (such as effective enquiries) Secondly, the form of the conversion funnel helps us strengthen our view of the process, but it itself weakens the segmentation of the process. This sounds contradictory, but it is not difficult to understand. The processes that can be constructed by funnels are all large processes and cannot include the user's more subtle behaviors. However, it is these more subtle user processes that reflect and even govern the occurrence of conversions. An extreme but very common example is a single page promotion. Conversion funnels rarely work on single-page promotions. Therefore, the conversion funnel is extremely valuable, but as I wrote in another article "Optimizing Conversions: In addition to the conversion funnel, you also need several analytical weapons in your arsenal", the conversion funnel is not sufficient to support all conversion analysis. However, we still have good weapons in our hands to solve this problem. This will be discussed in detail in the following methods. Method 4: Bubble analysis of core word interaction and ROI With this method, you first need to put different roots (core words) in different units - that is, subdivide all keywords according to the roots. But this is an ideal situation. Generally speaking, the same root word may be scattered in different units. It doesn’t matter. We have Excel. By using Excel’s table tool and then “Text Filter -> Contains”, we can easily filter out the relevant word roots, as shown in the figure below. For example, I want to find the core word with the root word "在职", enter "在职" in "Text Filter -> Contains", and then create a new column "Core Word 1: 在职", and fill in "在职" in column C corresponding to all the filtered results. Then, using a pivot table, you can summarize all the keywords according to the performance of their corresponding core words (roots). Extract the overall performance of each type of core word, and you get a core word performance list (CPL is Cost Per Lead, the lower the value, the better; green in the entire table represents good, and red represents bad): Now you can see that different root words have significant differences in the final results. The Leads/Click and CPL columns are very telling. Do you remember the ROI and engagement interaction model I talked about in class (the bubble chart)? There are many variations of this model, but the idea is simple: compare the interaction behavior of traffic and the conversions they ultimately generate. In most cases, good interaction (engagement) can correspond to good conversion (ROI), but there are always exceptions. These exceptions become our breakthrough points for analysis and optimization. We now need to use this model. The methods of this model are clear: We use the segmentation method of the website analysis tool (I won’t go into details here, you need to be proficient in using GA, basic skills), and summarize the traffic engagement performance of various words to form a more comprehensive data performance data, as shown in the following table: Using the data in the table above, we select CPL and Bounce Rate to make a bubble chart, as follows: Generally speaking, traffic with good engagement (lower bounce rate) will have better conversion capabilities (lower CPL), but as mentioned before, there are exceptions. The green line is the “normal performance” part, but outside the green line, there are two words with relatively low bounce rates and relatively high CPL, especially the word “tuition”. A comparable term to this is “Master of Finance”, which actually has a higher bounce rate than the tuition fee, but performs better in ROI (CPL). What is your strategy after this happens? Basically you have to do the following: 1. Check whether the creative idea is misleading; 2. If there is no problem with the creative idea, there must be something wrong with the landing page. A common situation is that there are elements on the page that attract users to click, but after users click, they find that the relevant information is far from their expectations - for example, the price is higher than they expected? At this point, you can continue to look at tools such as GA to see how the people behind this word perform after entering the landing page (for example, using the next page report I taught in class). The comparative analysis model of Engagement and ROI is one of the most convenient ways to analyze segmented traffic. Keyword analysis for paid search rankings is also useful. Method 5: Comparative analysis of core word interaction and ROI of single-page promotion This method is an extreme case of method four, but this extreme case is particularly common. For example, the following picture: Damn it. . . The bounce rate of all keywords is 95% or even higher. How can we survive in this situation? ! Is your delivery completely ineffective? In the cases I encountered, there was only one reason why this happened: the landing page where the keyword was located was a single-page promotion! The entire landing page of a single-page promotion cannot link to other pages. According to GA's definition of bounce rate, if the traffic cannot go elsewhere after it comes, then it must have bounced. So in this case, there is no need to look at the bounce rate. Therefore, the method 4 above is not applicable in this case because GA cannot provide effective engagement data. Either the bounce rate is close to 100%, or the dwell time is approximately 0... So, when we see that the two core words "Master of Finance" and "tuition" have similar costs and traffic, but have a huge difference in ROI, do we have any other tricks? (Let’s take a look at the picture of method four). At this time we need to use heat map. For example, using ptengine, there are contrast heatmap and segmentation heatmap functions. The traffic filtering and segmentation function is used here: in the filtering conditions, select all the words under a core word and segment the traffic of these words. The heat map on the left shows the traffic performance of all core words related to "tuition fees", and the heat map on the right shows the traffic performance of all core words related to "Master of Finance". The “tuition”-related traffic on the left shows more page views (the proportion of people who can read to the bottom of the page is significantly higher than that on the right), but the time they stay on the page is relatively short (the color is not as red as the one on the right). The “tuition” traffic seems to show a quick need to find information (willing to see the bottom of the page), but lacks better consultation (leads) conversion. Such data means that you should carefully check whether there is enough attractive "tuition" information on the page, or whether there is a lack of relevant information on this page at all. This data and heat map give us an important hint that the traffic desires behind these two core words are quite different. We must create landing pages with different information emphasis for these two words. Method 6: Correlation analysis between keywords The last method falls into the category of attribution. The most important application area is to see which other words searches the brand word searches are essentially derived from. There are many unknown brands that, in the early stages of SEM, have to expand their traffic range and use general words, competitor words , and crowd words. The purpose is, on the one hand, to attract traffic (but the conversion effect of traffic must be quite weak), and on the other hand, to expand brand exposure and prepare for future user conversions. As time goes by, people start searching for brand words. You must want to know which common words initially brought the traffic to these brand word searches. These traffic will be impressed after seeing the brand and will then search for your brand again. To be able to achieve this, we need to configure our monitoring tool. Take Google Analytics (GA) as an example: In order to view the attribution of keywords, GA has two prerequisites in its settings: 1. The website needs to set a goal, the setting location is as shown below: 2. Use the UTM link tag to accurately track keywords (detailed explanation in my course). 3. The relevant data report can be seen in GA's "Conversion->Multi- Channel Funnel->Popular Conversion Paths", but please note that you must select the "Keyword" item in the red box as shown in the figure below. Using GA (or other tools with similar functions), you can see the sequence of keywords searched by the same person - all the ones starting with "索" are traffic brought by brand words, and the word before it is a common word for placement. In this table, some common words were placed, which later led to searches for brand words. Therefore, your keyword optimization strategy changes from just looking at the conversion of brand words to not only looking at conversions, but also looking at how many brand word searches they can bring. Okay, I have explained the six methods. Which ones are useful to you? Any suggestions? Everyone is welcome to leave a message and discuss!Mobile application product promotion service: APP promotion service Qinggua Media advertising The author of this article @宋星 is compiled and published by (Qinggua Media). Please indicate the author information and source when reprinting! Site Map |
<<: Pulandian SEO Training: SEO Company | How do large websites perform SEO optimization?
>>: How much does a four-day tour to Hong Kong and Macau cost?
Recommend
Chen Longyu's Talisman Numerology Special Class 13 lessons + handouts
Chen Longyu's Talisman Numerology Special Cla...
User operation: Analysis of Didi’s user growth system!
1Functional module analysis and product introduct...
How to operate social e-commerce well?
We always say that when the wind blows, even pigs...
2019 Spring Festival Mobile App Market Insight Report!
1. In the past 30 days, the amount of material re...
PR software practice: 20 short video money-making lessons to teach you how to earn over 10,000 yuan a month
First of all, I would like to thank Liangmi Qingn...
Tips for acquiring customers through Weibo Fans Advertising!
It’s time to share with you our nearly 10 years o...
How to obtain seed users during cold start?
In my previous company, I was involved in product...
Ten marketing trends for 2022
At the end of 2019, there was a very popular joke...
Operational Tips丨5 secrets you don’t know about event operations!
Event operation is a job that all operations prof...
September marketing promotion hot calendar!
Autumn is the season of harvest. In the upcoming ...
Read the full text of The Widowed Sister-in-law novel for free, and read the full text of The Rural Widowed Sister-in-law for free without pop-ups!
Introduction to the novel of the widowed sister-i...
Zhihu, forced to live stream
On September 30, a Zhihu user said that he had re...
Hanjia Army 2021 First Quarter Practical Sharp Knife Class 8 Documents
Brief introduction to the document resources of H...
Practice: Full process analysis to improve APP Push conversion rate
Recently, we are working on optimizing the succes...
Check out the four growth strategies used by big companies like Tencent and NetEase
Behind the success of large companies, in additio...