? (Data monitoring period: 20 180507-0520) Xu Bowen 2018-11-30.
First, the research background and purpose, clear problems
background
The last three classes have been testing the teaching effect of "Internet Operation P 1 Series Course", which was launched in Baidu PC, Baidu Mobile, sogou and 360 for two weeks respectively. The test results are analyzed and evaluated.
purpose
1. Analyze the data according to the monitored data, calculate the delivery effect of each channel, and carry out quality transformation;
2. According to the data, according to the characteristics of different channels and different nodes, adjust the advertising strategy to achieve accurate advertising and improve the conversion rate /ROI.
Secondly, sort out data sources and user contact points.
User experience path
Users are moved by advertisements through various channels (keyword search, etc.). ) and began to enter official website in three classes.
The main process: enter the page → click register → login/registration (enter directly in the login state) → order confirmation → order payment → successful payment.
Detection indicator
Thirdly, the data funnel model is established
Step 3: According to the given funnel model, substitute the data and draw the map.
This test mainly monitors the relevant data of four nodes: landing page UV, registering page UV, successfully placing page UV, and successfully paying page UV.
According to the data, the conversion rate of the following key nodes can be calculated:
Registration conversion rate: registration page UV/ login page UV
Order conversion rate: order page UV/ registration page UV
Payment conversion rate: payment success page UV/ order page UV
Raw data:
All (4) funnel models of channel conversion rate
Funnel model of channel conversion rate
Fourth, data analysis and problem discovery.
The fourth step is to observe and compare the data of different channels, infer their operating conditions and clarify the main problems of each channel.
4 List of Channel Delivery Effect Data (20 180507-0520)
Note: The registration rate is marked on the auxiliary shaft.
Note: Red data is the lowest value of this series.
According to the chart, you can find the problem:
Baidu PC channel has the best delivery quality, and its registration rate, order placing rate and payment rate are higher than other channels. The ROI is nearly three times that of Baidu's mobile channel (0.34% and 0.09%), and the landing page UV is also at a high level.
Baidu's mobile channel has the worst delivery quality, and the landing page has the highest UV, but the registration conversion rate is very low, less than half that of Baidu PC, the order rate is also the lowest among the four channels, and the total transaction conversion rate is the lowest;
The channel delivery quality in sogou is at a low level, the ROI is less than half that of Baidu PC, and the user's landing page stays for the shortest time.
The quality of 360 search channel is poor, and the order rate and payment rate are far below the average level.
The consulting ratio of Baidu mobile terminal and Baidu PC terminal is higher than average.
The page bounce rate of sogou and 360 search is above average.
The data shows that there is room for optimization in registration rate, order rate and payment rate.
Law:
The user's stay time on the landing page is positively related to the proportion of consultation. The longer the stay time, the more the proportion of consultation.
The proportion of consultation is positively related to the payment rate. The more consultations, the higher the final payment rate.
The higher the user bounce rate, the lower the conversion rate.
Speculate:
Note: The verification of speculation is in the "five-year" analysis.
Is the low registration rate and order rate of Baidu Mobile related to the mobile platform? Will the page characteristics and interaction problems of the mobile terminal affect the experience: registration process, order process and payment process? (in the "five-year" analysis)
According to the five or six points in the "Question", the user portraits of users who use sogou and 360 search may be obviously different from those of Baidu's two channels. Different user types, user education level and urban distribution may affect the matching degree of users to the three categories (the coincidence degree with the three types of target users is not high), resulting in low stay time, high bounce rate and low consultation rate, which in turn leads to low conversion rate.
Is the low registration rate related to the place of admission? (Click on the registration place and watch it after 4-5 screens); Is it related to the long/complicated registration process?
The conversion rate of the order → payment process is low (44.7 1% on average, and more than half of users are lost in this process), so it is suggested to continue to monitor more data and further analyze the conversion rates of various payment channels; And how to continue to stimulate users to place orders on the order page, and what information to show? What copy is on display? It can give users positive feedback on continuing to pay.
Suggestion:
The fifth step is to explain what evidence or data are needed to prove/refute your hypothesis and propose a solution to the problem that is difficult to give a solution.
According to the test data of 14 days, Baidu PC channel has the highest ROI. It is suggested to increase the input resources of Baidu PC and reduce the input resources of sogou and 360.
Although the delivery quality of Baidu Mobile is the worst, don't directly characterize this channel as the channel with the lowest ROI. The channel payment rate and the consultation ratio of stay time are higher than the average level. It is suggested to analyze the reasons for the low registration rate and low order rate of Baidu Mobile. If the process and design affect the user loss caused by the experience, we can optimize it, and then further test it, monitor whether the optimized data improves the registration rate and order rate, and then decide whether to make a decision.
By comparing the portraits of three types of target users, we can find the regularity, which can reduce the overlap of the portraits of sogou 360 channel users.
Analyze whether sogou and 360 search channels are different from Baidu PC in terms of advertising guidance, page interaction (all three types are related to pages), registration process, order placing process and payment process, and whether there are any places that affect the experience and conversion rate. If so, whether it can be optimized and combined with the conclusion of recommendation 3 to decide to reduce the proportion of advertising resources.
Data chart of delivery effect of each channel
Note: The registration rate is marked on the auxiliary shaft.
Baidu PC:
Baidu mobile phone:
Sogou:
360 search:
Found the problem: the registration rate is too high, and there may be false data.
Channel 4 landing page UV
Found the problem:
The landing page UV fluctuates regularly with working days and weekends (5.12/135.19/20 is the weekend).
The fluctuation of PC and mobile terminal is different on weekdays and weekends.
Mobile (Baidu Mobile): On weekdays, UV fluctuations are stable and the data is low; UV rose sharply at the weekend, fluctuating greatly, and began to decrease near the working day.
PC (other 3 channels): the fluctuation of UV on working days is stable and the data is high; UV dropped sharply over the weekend, fluctuating greatly, and began to pick up near the working day.
4-channel registration conversion rate
Found the problem:
The registration rate of 360 search on June 5438+03/ 19/20 far exceeded the average level, so it is speculated that there may be false data.
Baidu PC fluctuated greatly on September 9 13 and 16, with a large increase.
Baidu Mobile has the best registration conversion rate, while Baidu Mobile has the worst.
4-channel order conversion rate
Found the problem:
The conversion rate of ordering on Sunday is relatively low (the monitoring period and data volume are not enough to support the qualitative analysis of the law, just guessing)
The conversion rates of Baidu PC and sogou are better, and Baidu Mobile is the worst.
4 channel payment conversion rate
Found the problem:
Baidu PC and Baidu Mobile have the best conversion rates, and 360 search is the worst.
According to the data, we can see that the payment conversion rate is the highest on Friday (118) (the monitoring period and data volume are not enough to characterize the law of data fluctuation, but only a guess).
Verb (abbreviation of verb) hypothesis and inference
Step 5: Make assumptions about at least two questions, and explain the main causes of the problems respectively.
Problem 1: The registration conversion rate of each channel is low.
Suppose 1: the deep entrance of the registration entrance (the "Register Now" button) leads to low conversion rate.
Inference: It takes about 4-5 screens to see the PC-side registration/registration entrance, and then further registration and ordering can be made.
Solution:
1. It is suggested to add a registration entry on the first screen;
2. The registration entrance hangs at the bottom of the page.
Hypothesis 2: Course introduction and copywriting need to be optimized.
Inference: The course introduction is not attractive because of the lack of syllabus.
Solution: Improve and optimize several angles:
1. What is this course and who is suitable for it;
2. The pain points of this group of people;
3. Why did you choose this course? What are our strengths?
4. Curriculum outline;
5. How to prove that the course is good (positive and side);
6. How awesome the tutor is;
7. We are different from other trainings;
8.7 days no reason to promise;
Question 2: The registration rate and order rate of Baidu Mobile are low.
Suppose 1: the problem of displaying page information on the mobile terminal affects the conversion rate.
Inference: The page size and resolution of the mobile terminal are not conducive to displaying the same information as the PC terminal, thus affecting the conversion rate.
Solution: Optimize the mobile terminal adaptation and information display, and optimize the registration portal. It is suggested to add a registration entrance on the first screen and a registration entrance suspended to the bottom.
Hypothesis 2: The mobile registration process is complicated/there are bugs.
Inference: The low registration rate of mobile terminals may be related to the complicated registration process or possible bugs. During the test registration, it was found that clicking the verification code input box did not trigger the numeric keypad, which led to the failure of registration and the loss of users.
Solution: reduce page jumps (try to complete registration at the bomb level) and optimize the registration process.
Question 3: sogou and 360 have low conversion rate, low consultation rate and high bounce rate.
Assume that 1: the user portrait has a low degree of coincidence with three types of target users.
Inference: It is inferred that the target user attributes of P 1 operated by Class Three are mostly women, aged 22-30, and 1/2 tier cities. The picture below shows the user portrait of Baidu index and 360 trend on keywords. It can be found that:
1. Gender: More women use Baidu than others, and 360 and sogou women account for a small proportion.
2. Age: Baidu is mostly 25-34,360, and sogou is mostly 19-24.
According to the above data, it is found that the users of Baidu channel have a high degree of coincidence with the three types of target users, so it can be speculated that the better delivery quality of Baidu channel may be related to the user type.
Solutions to intransitive verbs
The sixth step is to explain the reasons for clear and simple problems (give evidence) and give solutions;
According to the test data analysis of 14 days, Baidu PC channel has the highest ROI. It is suggested to increase the input resources of Baidu PC and reduce the input resources of sogou and 360.
According to the characteristics of UV fluctuation of PC and mobile terminal, it is suggested to increase the input resources of PC terminal on weekdays and mobile terminal on weekends.
note:
Although the delivery quality of Baidu Mobile is the worst, don't directly characterize this channel as the channel with the lowest ROI. The proportion of payment rate and stay time consultation in this channel is higher than the average level, and it is a good delivery channel with great potential. Suggestion:
1. First analyze the reasons for the low registration rate of Baidu Mobile. If it is determined that the process and design have affected the loss of users caused by experience, targeted optimization can be carried out.
2. After optimization, further online testing will be conducted to monitor whether the optimized data will improve the registration rate and order rate.
3. After drawing a conclusion, then decide whether to increase or decrease the resources invested in this channel.