Home big battle, what should I do? How to optimize the asset allocation of the homepage of the website (i)

Intermediary transaction SEO diagnosis Taobao guest Cloud host technology Hall

Objective】

My old colleague Xie always called me suddenly, asked me if their company's various departments are competing for the home page resources, and play the full time, how should coordinate the home page resources, to ensure that the company's overall business has the largest output. Lao Xie's company is a proprietary + platform E-commerce site, home is currently more available to the platform of Third-party businesses to obtain advertising revenue, but the proprietary sector has asked to reduce the third party on the home page of the invasion, and come up with a seemingly very convincing argument- The advertising costs of franchisees may not be as profitable as our proprietary goods, and they weaken our company's brand.

I suddenly realized that this is a very important issue, home page, a website façade, all departments have to contend for the interests of the focus, subject to political will does not produce organizational benefits maximization, whether there is a way to help us scientifically coordinate the will of all parties, and can produce persuasive results? In the company I was in , there will also be prolonged competition for the home page, but although the company does not have a platform and third-party businesses, but the internal departments are still in conflict with each other, quite tense. Homepage, it is impossible to mix content in the most optimized way, but to succumb to political compromise. I have said that the focus of the home page is about the result of political compromise.

On the phone, I explained to him that we have a number of ways to help determine the strategy of the home page, which can be explained in quantitative conclusions to help him convince stakeholders to accept reality. The phone for 30 minutes, we all know that only by phone can not tell the problem, so I promised to come down, I said, I should have an article.

Body】

Home page problem is difficult to solve the reasons for the following several aspects.

First of all, the home page is a collection of information, its role and goal of the target is actually more ambiguous than all other pages, which makes the optimization of the home page become somewhat "random." This means that before the allocation and optimization of home resources, we have to figure out what exactly we want to achieve on the home page what purpose? On this question, I asked him carefully, I ask him, do you particularly care about the company's brand image, the company's sales of products (whether high-end or at least not low-end), Or do you need to highlight your industry-leading standards and authority? If the answer is "yes", then, we simply use some "efficacy" indicators to measure the home page is no longer appropriate, but must take into account the "public welfare" value of the home page. This is similar to the difference between "effect marketing" and "brand marketing" when we're doing internet marketing--If your goal is branding, it's understandable that you're not driven by sales or conversion effects, or even at the expense of a significant portion of the transformation effect. In this case, you may feel that "political" factors in your home page, and optimize the home page goal, process and results are no longer so "quantitative", but the goal is so, there is nothing to think about. However, if the goal of the home page is to maximize the current economic profits, things will be much better, we can clearly define the metrics to be used, methodically carry out our tests, and give sufficient proof of results and impact factors analysis and conclusions.

Therefore, in the following discussion, we all assume that the goal of the home page is "utilitarian", that is, in order to achieve the current economic benefits of maximizing.

General method: Who is more productive?

The general method of solving the homepage problem will not exceed everyone's expectation. The best way to quell the controversy is to "be a mule or a horse." For the old Xie Company's problem, it seems to be able to promise: "We will continue this week with Third-party ads on the home page, and the next one weeks, we will be placed in the internal proprietary Department of the product recommendations." "Two weeks later, we look at the data and talk.

Our last look at the performance data is of course the revenue data (for the third party platform department is the revenue of marketing advertising, for the proprietary sector is the sales margin). If your situation is different from old Xie, and similar to my previous company-there is no Third-party platform business, and all is self, the competition is the home page of different categories of competition-then the performance data comparison is simpler, that is, compared to different departments in the specified time after the home debut, who contributed more Maori.

I believe most business friends can think of this way, this method is intuitive, easy to operate, and seems easy to convince people to accept.

Wait!

If you are really prepared to adopt this approach, I fear that in the end it will be ten to one.

The problem is not whether we set output as a relative key point. Generally speaking, this setting is correct and understandable. The problem is that comparing the time required to set up different home pages is a big practical hurdle. In general, it takes a week for the data to accumulate, and two different tests at least two weeks-two weeks, with no such patience in the business sector. Even with patience, time is still a problem, because no one can guarantee that the environment will not change during this time, for example, in the second week when the company suddenly has a full range of discount promotions, then the second week's performance and the first week can not be compared. In short, the time has changed, the environment will certainly change, this is not Apple to Apple's comparison, certainly can not achieve your desired fair comparison effect.

So what can be done to help us get a fair and effective comparison?

A/b test is better than time-sharing test

We have to conduct a/b test. If possible, all comparisons of internet marketing should be tested with AB tests. A/b test avoids the awkwardness of the above time-sharing test--ab test is simultaneous (eliminates the environmental changes caused by time), the sample is random (reduces the sample deviation), the process is controllable (I can stop at any time), and the result can be reflected in real time. All this is destined to make this method of great significance.

E-commerce companies that often do A/B testing often have higher operating levels. For what is a/b test, please take a look at this article: avinash– Analysis Series 2: Experiment and test initiation.

If you use Google Analytics, it has come with A/B testing tool (the integration of the previously independent Google Website Optimizer tool), in comparison with the output efficiency we would like to see, simply set to achieve our needs. For example, the following settings, "Transactions" refers to the two of your test home to bring the number of transactions, and 25% refers to the time you have to do the test of the flow of traffic to reach the home page of the total flow of 25%, that is, two of the first page of the flow of 12.5% and 12.5%-- In fact, one of the home page must have 87.5% traffic, but A/b test tool statistics only randomly selected 12.5% of them. Obviously, the smaller the ratio is, the more secure it is, but the longer it takes to accumulate the data.

A/b test is almost the only way to finally quell the quarrel. But A/b test requires a new page, a bit cumbersome. If you only need to test some of the content on the home page, such as just the top of the page (leaderboard) is the focus of the quarrel, then some friends will consider the choice of multivariable testing. This approach is very similar to A/b test, but does not require the design of multiple pages, but only the partial division of the page into a, B version to replace-the two ideas are exactly the same. For the Home page optimization, multivariable testing does not appear to add pages, save some operations, seems to be more efficient, but the trouble is also here, because you will know that the special add a B page in fact there is a big use. In addition, multivariable testing tools are almost always paid.

The world's largest paid web analytics testing tool is optimizely and Adobe Test & Target, which are almost equivalent, and are not easy to use. Quickly deploy the way you implement your tests, which is better suited to leveraging Google Analytics experiments tools.

AB and multivariable tests were originally designed to achieve a single metric, that is, I bet you that a home page can bring a better conversion rate than the B home page, so the conversion rate is the only indicator of this gamble (test). However, our goal is to optimize the home page of the asset allocation, we often have to take into account other performance, so the purpose is necessarily not so simple. Let's go down and see how we can solve the problem.

Pointers to other important behaviors other than outputs

If you just follow a tool's module to do a test with only one metric, it's a bit of a waste to test this "cow x" approach. The most valuable place for A/B testing is not that it gives the game a result, but more importantly, it can randomly allocate traffic to two (or more) of the same purpose and design different pages. All the user behavior of these two pages can be faithfully recorded by Web analytics tools for your detailed analysis. It's wonderful! (So, in fact, even without a A/b test tool, as long as your site's front-end engineers are willing to help you achieve the flow of traffic randomly on the two home page, you have a Web analytics tool, you can make a very reliable A/b test.) )

Now that you have these two weapons--a/b test tools for automated random streaming and web analytics tools, what else can you do?

The analysis of key performance indicators (KPIs) on the home page of different asset configurations can satisfy our curiosity to a large extent, and let us know how the home page of different resource configurations affects the behavior of site visitors. These KPIs include the bounce rate (bounce rate), the test area clicks (CTR), the click Distribution (click on the hot Chart), and page value. But I did not put the page to include in the KPI, because the length of stay on the front page is difficult to explain what, if a page stays longer than the B page stay time on average more than 5 seconds, perhaps only a reasonable range of data itself disturbances. The home A/b test reflects the difference between the average stay time I have never encountered.

Of the above KPIs, I think the most important thing is still the bounce rate (bounce rate). If a home page bounce rate than the B home more than 10%, we have a good reason to suspect that the effect of a home page really reduced. Even if the final a-home traffic conversion rate than the B home high (this inverse of the day is very rare), we can not say that this is a better than B, but only said, may be a home page recommended the category of goods more likely to persuade some people to pay for the purchase, and because of this part of the promotion and the overall conversion rate was promoted.

Give me a simplistic example. Suppose the top of an ecommerce site has a recommended product display for each department, in order to quell the controversy, the website department conducted a A/b test. The establishment of a, B page is only recommended to recommend a different product. Test results are shown in the following table: A home page bounce rate is the 60%,b home page is 50%. Suppose two pages are greeted with 1,000 visits, and the odds of buying other bounce are the same for visit that have not been lost. A page has 600 access bounce off, the remaining 400 of the 50 purchased a home page recommended products, 20 purchased other goods, while the B page has 500 access bounce, the remaining 500 of 30 of the purchase of B home recommended products, 25 bought other goods. The final conversion rate for page A is 7%, while the B page is 5.5%.

This situation shows that a home page recommended products for those who like this kind of goods is too good to sell (400 of 80 people buy), but others may be because of the impact of the product on the home page interest (bounce rate of 60%). A home better? or bad? The conclusion depends entirely on your own criteria (such as adding other key indicators, such as sales or profits). But if we just use Google Analytics's experiments function to do a A/b test and set the conversion rate as a measure, then we can only conclude that crystallizes home is better than B home. But the fact is, a home page is not necessarily better than the B home, although it recommended the goods are more popular with some people, but it reduced the overall performance of the page, and affected other sales opportunities.

So I'm not particularly fond of the simple test of a wager. I'd like to know what kind of behavioral differences two different pages are causing and why.

The change in the click rate of the test area (the changed area on the home page) is the most intuitive indicator of how people change their behavior as the home page changes. This click-through definition is: Home page The number of times the link is clicked (in fact, the first page of the area link is clicked after the PV number of pages opened) divided by the first number of visits. After a region changes, if its own hit rate significantly increased, and other regions have not significantly reduced CTR, it is clear that this change is valuable. In the following example, the BCD three new home page to do the original first modification, the same area, the content is different, the click rate is also different, and the link on the left of the click-through actually did not occur significantly changes, it is obvious that the D-home effect is the best. This example four version actually sells is a kind of commodity, but we can imagine that it promotes separately four different department's goods, each department's performance is various, finally the D department wins the attention force to win (the test area's click Rate is highest).

However, if there is a significant adjustment to the home page as a whole, the individual test area will no longer exist. At this point we are more commonly used to judge the interest of people by analyzing the two hits on the home page. The following site is a UK job Description professional website, home page in a quarrel set to the second edition. Proponents of the original version (a) argue that visitors need to know more about the relevant job requirements and other skills and experiences that can help them learn how to find a job, so it is necessary to place more information on the page. Supporters of the second edition (version B) objected to the idea that they wanted a more pure homepage, that is, to help people find jobs with their professional counterparts, and that the more detailed classification and information be done by clicking on the relevant categories or pages after the navigation.

Let's take a look at what the version A and B are all about here:

A version of the first page is as follows:

b version of the first page is as follows:

Now, the two page is placed on the AB test stage, waiting for the audience to vote. I do not want to give you suspense, the final conclusion is the B version of the page indicators (a detailed explanation of indicators read this article: http://www.chinawebanalytics.cn/ web-marketing-key-metrics-and-logic-2/) to win over a version 21%. But after the original version changed to the second edition, what happened to the user's behavior? The following click Contrast diagram shows some of the phenomena. All numbers for the assumption that two pages have 1,000 access, the corresponding area of the number of clicks received.

A page of visitors seems to be slightly more dependent on navigation to find what they want, a page of navigation click More than B. A page in the job search category (above the page of the accouting, Banking, Consulting, and more of the five categories) poured more detail, but less than the number of clicks on the B page-this shows that more information does not necessarily get more clicks. Limited space information is always very difficult, not put on the information is easy to be mistaken for users of this information, resulting in the user instead of giving up click to see exactly; on the contrary, only to give large classification and attract the eyes of the big picture, but more likely to trigger the user to explore the desire. A page of related downloads (26 clicks on the left page) is not appreciated on the homepage, but in fact these downloads are one of the most popular content on the site-home page function of multiple distracting people's attention, but also ignore the people receive information in order, People may not need to download the first page when they have not yet returned to God.

b home So get a better click effect, the total number of 1,000 hits is 776, more than 639 clicks a home page 21%. The focus and streamlining of information and visual elements on people's mentality and action are clearly shown on the hot zone map.

"To be Continued ..."

In the upper part, the focus is to explain what is the most appropriate method of data determination to determine what the home page is more suitable for the organization's pursuit of the effect (AB test and multivariable test). And, in addition to focus on the final indicators, we should also pay attention to which user behavior pointers to further help us analyze when the home page asset allocation changes, the home page utility rises or decreases. In the second part, you will see some of the more "high-end flavor" approach, as well as some other optimization of the home page common solutions and their effects. Please look forward to it! Now the article is less, but promise everyone, or not write, to write on dry goods!

Original: http://www.chinawebanalytics.cn/homepage-fighting-what-i-should-do/