Traffic model and homepage indicator system: Homepage strategy operation

During the long Spring Festival holiday, everyone read a lot of books and articles, right? During this special period, I hope all my readers are safe, healthy, happy and peaceful, and have a fulfilling and rewarding life, without having to share nuts out of boredom.

Although I haven't posted an article for a few weeks, I have not been idle at all. Whether I was on a plane to or from the United States or during the Spring Festival, I have been busy writing. Finally, I have finished writing the sister article of the homepage product strategy that many readers have been waiting for - the homepage strategy operation article. I hope to bring you a feast of practical knowledge. The content is quite comprehensive, more than 20,000 words, so I will publish it in three parts.

Internet operations cover a wide range. In addition to the four basic operation systems such as activities, products, users, and content, there are also new media operations, traffic operations , category operations, store operations, game operations, points operations, and other operations that focus on specific goals, as well as product operations that are intertwined with product functions.

The homepage of the entire site is undoubtedly the leader of all pages, and the operation of the homepage is the top priority of operations. But along the way, few books or articles on operations mentioned how to operate the homepage well, improve user experience and traffic efficiency, and maximize business output. This article shares the overall framework of homepage operations based on my many years of practical experience in charge of product, operations, and data teams in large e-commerce platforms. The main ideas in this article also apply to channel homepages and store homepages.

This article is for e-commerce apps. The homepage designs of apps of different business types may be somewhat different, and this can be extrapolated across industries.

The content framework is as follows:

1. Home page traffic model

To operate the home page well, you should first understand the home page traffic model, that is, where the traffic comes from, where it goes, and how it is distributed on the home page. Let’s talk about the origins and characteristics of home page traffic.

1. Source

Home page traffic usually comes from the following sources:

1. Direct access: Users open the app on their mobile phone and go to the homepage.
2. Promotion/Sharing - Deeplink Jump: embed deeplink code on SEO/SEM web pages and social sharing pages, specify the URL, and the user opens the web page with a browser or WeChat, and uses deep link technology to call up the app (if it is installed) and go directly to the corresponding page.

Call the app via Deeplink in the browser

For WeChat to invoke, the app needs to be on Tencent’s whitelist (mainly Tencent’s relatives, such as JD.com, Ctrip, etc.). Non-whitelisted companies are only allowed to open web pages in WeChat and cannot invoke the app. At this time, you can guide users to open it through a browser that supports deep linking, and then jump to the app on the browser.

WeChat wake up application

3) Third-party app call-up: Through cross-traffic diversion through cooperating applications, the app is called up in other applications and directly reaches the designated page, such as jumping from Douyin to Taobao. This is technically feasible, but in most cases, traffic management rules should be established, and traffic should not be easily allowed to jump out of the app and "pick a hole in the rice bag."

4) System shortcut function entrance: negative one screen supported by most mobile phones such as iPhone. Swipe right from the first screen of the phone, and the page on the left side of the first screen you enter is the negative one screen, where you can place components to point to functions within the app. In addition, there is 3D touch, which has been cancelled in the latest version of the iPhone. This is a feature that was supported starting from the iPhone 6S, but users ultimately failed to develop a habit of it. It displays a set of specific entrances by pressing the app icon hard, and clicks to go directly to the specific page of the app.

Negative one screen

3D Touch

We see that not all app traffic must go through the home page. Direct visits (directly opening the app) usually account for the majority of app homepage traffic.

2. Where to go

After entering the homepage, users generally go to the following places (data is not deduplicated, and the proportions are for reference only):

First screen traffic distribution example

1) Search traffic: Users with clear goals directly enter the search box to search for the products they need. I usually call this part of the traffic precise traffic. There are also cases where you enter broad words such as "snacks" to match multiple subcategories, or go to the "category" entrance at the bottom of the column. This can be considered semi-precise.

2) Idle traffic: After the precise traffic is diverted, the remaining traffic flows down along the homepage and becomes idle traffic. Some users have clear preference columns, such as flash sales, and will quickly find the entrance by swiping on the homepage.

For columns that are located deeper, one idea for traffic flow design is to set up a floating elevator navigation at the top, which can scroll down directly to the corresponding area after clicking. Another idea is to force the traffic flow, where users can only swipe to the column screen by screen. Other columns along the way may also attract users to generate additional clicks and sales, but there is a risk of loss. These two ideas have their own emphasis. If conditions permit, you can do AB testing to make a choice.

As more users browse down the homepage, the ability of the column entrance to capture traffic is very important. The strategy is discussed in the homepage operation skills section.

3) Functional area traffic: Many e-commerce companies place functional entrances such as categories, content, shopping carts and personal centers in the bottom column. "Classification" provides a tree-like category navigation to help users find products in a hierarchical manner, while "content" (such as JD.com's Discovery and Taobao's Weitao) is an important way to plant grass and provide stickiness. These two entrances are more like shopping guides, which are qualitatively different from the shopping cart and personal center. Due to its prime position, content entrances often have good traffic, but non-content e-commerce conversions are generally poor. You may consider making some replacements during specific stages such as big promotions. In addition, the letter/message entrance within the site also belongs to the functional area traffic.

3. Home page traffic distribution

The home page traffic is undoubtedly 100% on the first screen. Then some traffic without a second jump (nothing is clicked) will leave directly. Part of the remaining traffic is diverted to the search box, and the rest is distributed in an inverted pyramid, decreasing from top to bottom, screen by screen, like this.

Traffic decreases screen by screen

Chinese people like to shop, and e-commerce companies have relatively strong shopping guide and operational capabilities. Shopping traffic often accounts for more than 60% of the total, while Americans like to search. According to Amazon data, more than 80% of traffic in the United States goes to search. This feature determines that Chinese and Western e-commerce apps have relatively different homepage design ideas. In the United States, most e-commerce companies do not even have operations positions.

Traffic reaches the bottom of the homepage, which is like reaching the end of a shopping mall. The flow at this time is called the flow residual value (in addition, the flow after the settlement process is completed can also be called residual value). Some apps will prompt "I have a bottom line" and then prohibit you from swiping further, which is a bit of a waste of data. Since users are still not satisfied, we can create a waterfall flow and make personalized recommendations so that users can continue browsing, just like watching TikTok. In recent years, people have become more and more fond of this worry-free way of strolling. This reflects the gradual change in the shopping guide concept from "people looking for goods" to "goods looking for people", and is also an important reason why JD.com and Taobao have greatly enhanced "Guess You Like" services.

The “Guess You Like” function of mainstream e-commerce accounts for approximately 5% to 20% of traffic. From the first screen to the bottom, the traffic distribution in the middle can be used to count the number of times each column is displayed. According to experience, more than 60% of the traffic only reaches the first and second screens, and then there is a cliff. The traffic after the third screen decreases at different rates depending on the column situation. Of course, exposure and clicks are not necessarily proportional.

The click ratio of the home page column reflects the proportion of traffic entering from the home page. Further statistics are made on the proportion of the column's sales to the sales on the homepage. The difference between the two reflects the column's sales capacity per unit of traffic. When I manage homepage resources, it will be used as an important indicator to evaluate the column's traffic usage.

When the traffic reaches three screens or deeper, you can place a "return to top" floating control to give the golden column on the first screen a chance to be visited again.

2. Home page efficiency monitoring and resource management

1. Home page operation core indicator system

The operation of the homepage must be targeted and have clear objectives. We cannot simply talk about whether it is good or bad. Therefore, the first thing to do is to clarify the North Star indicator of operations. Since the homepage is the entrance to almost all of the company’s businesses, a metric system can be set up. In addition to the most important North Star, a set of multi-dimensional indicators can also be used to comprehensively evaluate the efficiency of the homepage. These indicators are collectively referred to as the core indicators of the homepage in this article, and the North Star indicators can be selected from them based on strategic priorities.

Common homepage core indicators are as follows:

1) Revenue indicators. This group of indicators reflects the contribution of the homepage to revenue. Common indicators include the following three:

• Home page revenue: The amount of transactions that start from the home page and end in a certain period of time is counted as home page revenue (GMV, Gross Merchandise Volume). This includes both direct revenue and indirect revenue from the homepage.
• Direct revenue from home page: the amount of transactions completed directly on the home page during a certain period of time. Here the homepage directly completes the conversion. With the development of the product waterfall column on the homepage, this indicator will continue to improve.
• Home page click value: After calculating the home page revenue, calculate the total home page clicks (Click) in the same time period, then the home page click value = GMV/Click.

Note: Among these three indicators, the first two mainly measure the revenue share and direct contribution of the homepage, and the third mainly measures the efficiency of the homepage in guiding transactions. For large e-commerce companies or companies with extensive business, the main mission of the homepage is not to directly complete sales, but for store homepages or small businesses, direct transactions on the homepage can serve as an important indicator.

2) Interest indicators. This set of indicators reflects the attractiveness of the homepage to users and is calculated within the statistical period.

Home page second jump rate: What percentage of users jump a second time (i.e. click on a column, product, etc.) after arriving at the home page. Or conversely, how many users leave without doing anything after arriving at the homepage is called the homepage bounce rate. This indicator is more affected by traffic quality and crawlers, but it is usually monitored as an important indicator and is also the first link in the big conversion funnel.

• Number of home page visits: the average number of screens visited by users.
• Home page clicks: The total number of clicks by users on the home page.
• Home page visit depth: the number of PVs generated by home page clicks.
• Home page dwell time: The average length of time users spend on the home page. I think this indicator is relatively useless among the homepage indicators and does not directly reflect the value of the homepage, but it is often also a statistical indicator.
• Home page visit frequency: the average number of times users visit the home page during the statistical period. Except for days or smaller units, other periods should be deduplicated on a daily basis. There are often other reasons for users' high frequency of visits, such as setting up sticky channels such as check-in, new products, and flash sales, or the main categories themselves have high-frequency characteristics (fresh food, daily necessities, etc.). This indicator does not fully reflect the frequency driving ability of the homepage itself.

3) Guiding indicators. These indicators look at the homepage's ability to guide users to the target/core business.

Category penetration guidance indicator: Statistics on how many customers have made cross-category visits/purchases through the guidance of the homepage during a certain period of time. The statistical unit is usually the number or percentage of new customers who purchase across categories.

Specific business/category guidance indicators: When a company launches a new business or makes a strategic transformation to a certain business, it hopes to guide more non-business customers to this business (for example, Amazon China strategically guided Kindle users to the overseas shopping business at a certain stage). Since the homepage plays an extremely important role in guidance and traffic distribution, this indicator can be used to count the homepage's ability to bring customers to the strategic business. The statistical units are the same as above, and this can be considered a special case of the category penetration guidance indicator.

Traffic distribution accuracy. The accuracy of the distribution of the planned traffic demand ratio on the statistical homepage can be given by statistical methods.

In practice, the three groups of indicators, revenue, interest, and guidance, are highly correlated with the homepage product and operation levels, and a homepage indicator system can be built on this basis. Once the core indicators are clarified, the core direction of operations will also be very clear.

2. Home page management indicator system

An important responsibility of home page operations is to ensure the rational allocation and efficient use of each home page resource position. To this end, it is necessary to set up a homepage management indicator system to measure the efficiency of each homepage resource position. The following indicators can be used to build a homepage management indicator system.

1) CTR (Click Through Rate)

CTR = number of entrance clicks / number of entrance impressions.

CTR reflects the attractiveness of the entrance to users who see it.

One misunderstanding that needs attention is that CTR does not reflect the traffic share, it is highly affected by the entry location. For example, CTR=30% does not mean that 30% of users on the homepage click on this entrance. Among the users who only saw this entrance on the homepage, 30% entered the column.

For the entrance at the end of the homepage, the denominator is relatively small, and users who reach the bottom often have a stronger desire to "browse", resulting in an increase in the CTR of the entrance at the bottom, but the actual number of clicks is even smaller. Therefore, when comparing CTR, you should compare entrances that are close to each other, rather than comparing them globally.

2) Click Mix

Click Mix = total number of clicks on a channel entrance / total number of clicks on the homepage

Compared with CTR, Click Mix can better reflect the proportion of users entering the column. I usually use this indicator to judge the traffic acquisition of this column.

3) GMV Mix

GMV Mix = Channel Sales / Total Site Sales

There are many ways to calculate e-commerce revenue, such as order amount, transaction amount, and confirmed receipt amount. The main difference is whether to include canceled orders, returns, supplier rebates, etc. There is generally a rough proportional relationship between them. When calculating, make sure to use a unified caliber to calculate column sales and total site sales. It is recommended to use the transaction volume caliber – GMV.

Considering that the average prices of items in different categories vary greatly, Unit Mix (i.e. the number of items sold) can also be used here to eliminate the differences in the average prices of items in different categories. There is no absolutely fair algorithm. When calculated using GMV Mix, categories with high average unit prices, such as digital home appliances, have an advantage; when calculated using Unit Mix, categories with low average unit prices, large sales volumes, and high repurchase frequencies, such as supermarket products, have an advantage. It is also possible to calculate these two indicators simultaneously and use them together as a reference.

4) Conversion efficiency

Conversion efficiency = GMV Mix/Click Mix

This indicator reflects the conversion efficiency of the column. For example, if a column receives 10% of clicks and contributes 15% of revenue, the ratio is 1.5, which means that the conversion efficiency of the column is 1.5 times that of the entire site. Similarly, you can change GMV Mix to Unit Mix and look at the conversion efficiency from the number of sales.

5) Attributed GMV

Attribute GMV = GMV/access path depth

Let’s talk about the attribution algorithm first. Suppose a consumer visits channels A, B, and C, sees a certain product, and finally adds the product to the cart on channel C and completes the purchase. From the perspective of the attribution model, there are multiple options. I tend to adopt the "final channel interaction" model, which means that the credit is entirely attributed to the C channel, and whoever completes the final blow gets it. However, some analysts prefer to use linear attribution, which is equal distribution of A, B, and C, which is the algorithm of this indicator. A more complex and reasonable approach is to use a time decay model or a Markov attribution model. Students who are interested can check the relevant information.

6) GV Mix

GV Mix = the number of visits to the product detail pages generated by a column / the total number of visits to the product detail pages of the entire site

Here, GV represents the total number of visits to a product details page, that is, the product details page PV.

Most channels and activity pages display and recommend products through various dimensions (promotions, selections, new products, etc.) until the user reaches the product details page, where the goods are placed in front of the customer to guide the task to completion.

Whether the goods and prices can attract users to add to their carts, and whether the payment method can allow users to complete the settlement, is not the responsibility of the shopping guide channel. Therefore, the proportion of product detail page PV generated by the channel highly reflects the shopping guide value and importance of the channel.

7) Average PV per person

Average PV per person = Total PV in the channel / Total number of visitors to the channel

This indicator reflects whether users can browse the channel, that is, whether the channel is rich, exciting, and attractive.

8) GV per capita

GV per capita = total GV in the channel / total number of channel visitors. This indicator is a combination of the previous two indicators, reflecting the average number of single product pages brought to each user by the channel. If the conversion rate of the single product page is constant, then this indicator directly determines the conversion ability of the shopping guide column. Therefore, in practice, I usually use this indicator as the North Star to measure the effectiveness of each homepage shopping guide channel.

9) New customer conversion

New customer conversion = the number of customers who complete the process of adding a cart and finally place an order within the channel, changing from 0 to 1 order.

The channel is not just about selling goods, it also has to carry out high-value tasks, such as attracting new customers (converting new customers), selling memberships, and increasing stickiness. This indicator reflects the channel's ability to convert visitors from "browsing" to customers.

This indicator can also be expanded from new customers to new customers in the category. For example, a user who has never purchased beauty products before may contribute to category penetration by adding beauty products to his cart and ultimately purchasing them in this channel.

10) Member Engagement

Member engagement = total number of members who add items to the channel and ultimately place an order / total number of members who place orders on the entire site. Many e-commerce companies attach great importance to improving member activity. Whether a channel can effectively guide members to place orders and thus increase member engagement and activity reflects the channel's attractiveness to members, thereby increasing member conversion and renewal rates.

11) Channel visit frequency

Channel access frequency = channel non-deduplication access times / channel deduplication access times

This indicator reflects the stickiness of a channel to users and also reflects whether users who visit the channel are willing to visit again. Some channels focus on stickiness, such as content, short videos, and games. It is not reasonable to directly evaluate sales indicators for these channels. For such channels, the frequency of use indicator can show its stickiness to users, thereby making an indirect contribution to increasing the channel's traffic and sales.

12) Channel dropout rate

Channel bounce rate = number of people who leave the app on the page in the channel / total number of people who visit the channel

By comparing horizontally, we can see which channels cause more users to leave, which to some extent also reflects whether there are major defects (such as external links, freezing, crashes) or experience problems in the channel.

The above introduces 12 common indicators. You can also define corresponding indicators based on your own business needs and characteristics to measure the contribution of a homepage channel to core business goals.

3. Home page data report

The data reports that the operation system needs to track can be divided into two categories:

1) Home page overall indicators

The first category is the overall indicators of the homepage. A group of indicators are selected from the aforementioned core indicator system of the homepage, and statistics are conducted on a weekly basis. At the same time, the week-on-week and year-on-year comparisons are observed. If there are large fluctuations, in-depth analysis is conducted to find the reasons. In addition, you can also observe the monthly trend and analyze the continuous changes in the same trend according to the seven-point principle.

The report can refer to the following design, or you can define it yourself based on indicators and time periods.

Home page overall indicator example

We can see that the overall effect of the homepage is evaluated from multiple dimensions, and sometimes we will see that the performance of some dimensions improves, while the performance of some dimensions decreases. There is no need to pursue better performance in every aspect. You only need to ensure that the key goals perform well and continue to improve. At the same time, when the key goals are guaranteed, consider optimizing and improving the secondary goals.

2) Home page entry efficiency index

The second category is the entry efficiency indicator, which evaluates the performance of each resource position on the homepage. You can select a group of indicators of interest from the aforementioned homepage management indicator system, and also perform statistics on a weekly basis to observe the weekly and year-on-year (if the same channel occupies the same entrance) fluctuations and continuous change trends of each resource position efficiency, and perform analysis.

We can further establish a comprehensive resource position scoring model, take weighted average of key indicators, and calculate the comprehensive score of each channel. For example,

Resource position score = average GV score*50%+conversion efficiency score*30%+category penetration score*10%+channel bounce rate score*10%

In this example, per capita GV, conversion efficiency, category penetration, and channel bounce rate were selected as the resource management indicator system, with a ratio of 5:3:1:1.

Based on the comprehensive score of the channel, the following three management methods can be considered:

a) Cancel the entry of low-score channels and add an entry for high-score channels;

b) Split the traffic of low-score channels, retain part of the traffic, and allocate part of the traffic to high-score channels.

c) The entrance of low-score channels sinks, and the entrance of high-score channels rises;

Simply put, it is to use traffic to reward and punish.

The report can refer to the following design, or you can define it yourself based on indicators and time periods.

Home Channel Efficiency Indicator Example

The methods and specific cases for applying the management indicator system will be introduced in detail in the subsequent part of the home page strategy operation section, so stay tuned.

Author: Product meets operation

Source: Product meets operation