Want to learn about website operations? I'll teach you! Under

Transcend Dry:

In the previous chapter, we discussed how to choose and grasp the content. In fact, this is not the real secret of website operation. If you really want to win the market in the future, you must master A/B testing. This is the last chapter of this article and the most important part.

Some of these strategies are so important that you have to start right now. If you want to summarize, then look at the following paragraph:

If your site is loading and responding very slowly, you're losing your audience day after day, and if your site doesn't seamlessly switch between mobile phones and tablets, now let go of everything on hand, and fix it, If your home page does not produce some content with images and videos every one hours, your traffic is being taken away from your competitors. The urgent task is to deal with these problems.

In addition to these proven issues, you may want to do more to increase your brand loyalty. If a website is willing to experiment on the web by itself, to verify or overturn the assumptions that you have made before, to learn all the things that are necessary to learn, then it will be great. Please remember that only the data will speak.

The importance and feasibility of A/B testing

It is because of the experiments on the Internet that these internet giants are now able to head start. Google researchers said: "For every change, they have done countless experiments to verify how the user experience will change." This way of exploring yourself a little bit deeper, then the page is not just the choice between two options, A/B testing will be transformed into MVT (multivariate testing), that is, a page to render multiple variables, together to test.

Although companies such as Amazon.com and Yahoo have been taking AB tests since the 90 's, most news organizations have so far lacked the infrastructure or the lack of expertise to launch these internet experiments.

In fact, these experiments are not so easy to do. First, you have to keep track of individual users through effective means, and to cross-platform. If the user is not effectively and scientifically classified into different groups within the study, then the significance of the experiment can not be said.

Second, the new company needs to be able to provide several versions of the Web page. At present, the vast majority of newspapers do not have the ability to do this thing. But the problem is well solved. Cloud computing platforms, such as Amazon Web Services and Google App Engine, are very easy-to-use and very inexpensive tools. Of course, companies also need to verify that load times and response times behave the same on different servers. Now many companies provide A/B testing just to add a few lines of code on the target page, in fact, there are far higher than this, but also more scientific, and no cost of services. The new open-source multivariable test platform, such as Facebook's recently released Planout feature, is very powerful and can hardly cost the company much.

Gradually, the news company has no excuse not to do these tests. In fact, there are a lot of big media companies have started to do this work, and even opened a new department, such as Yahoo. But even so, they are still not very familiar with the direction of the site's innovation. You can see the words "Innovation, innovation and innovation", as well as the slogan of "Experiment, experiment and experiment".

In journalism, the discussion around AB testing often does not depend on whether the total flow is up or down, which in fact reflects the outdated media thinking: the majority of the audience is stable, any changes to the site will bring immediate changes from the total amount. In fact, the situation is not this way, you should first think that the Internet traffic is dynamic, only the first with this concept, A/B testing will make sense. The aim of the company's testing is not to evaluate the results from the overall flow, but to examine the growth rate. The real impetus of innovation and change will allow people to continue to return to the site, and constantly read more extension content. The test takes several weeks, even months, to evaluate the effect.

Industry-wide collaboration is an ideal way to achieve A/B testing

AB testing is necessary, but the cost is huge for media companies. In order to fully implement this test, you can choose to spread the cost through cooperation within the industry.

Online is particularly burdensome for small news organizations, and if costs are shared with a few readers, it is a number that sucks air. Large news media organizations such as time or Guadian have a relatively small cost to each person because of their large number of paying customers.

Big companies such as Google and Yahoo tend to perform thousands of of small tests, offering a variety of options and possibilities for future improvements. These changes based on test results can be very small, smaller than the color theme on the site, or the border width of the pixel. Although these changes are insignificant, the impact of the aggregation is far-reaching. Every bit of detail on the website of a large technology company represented by Google is the result of a trial that the user feels.

Newspapers, especially small-circulation newspapers, do not have the strength to do such network testing in the future. Aware of the changes caused by small changes not only because of the technical level of the smaller sites do not, and because the scope of research objects are limited, can not really reflect the site details of the impact of the changes. At this point, small news organizations and companies should be more likely to unite to do online testing, this time not only for low-cost reasons, but also through the sharing of talent resources in the industry, as well as the respondents together to produce meaningful results of the experiment samples.

Of course, some informal cooperation has now begun. A lot of the ideas and research you can see on Twitter and blogs. The American Press Institute (American College of Journalism), the Knight Foundation, and the Pew Research Center (Independent Polling Institute: Pew Research Center), as well as several other industry organizations and academic The heart has been combined to conduct a full range of network testing for news audiences.

These tests are just a few informal collaborations. The news media now needs a platform on which to customize the most basic experimental frameworks and schedules, share experimental results, get peer feedback, and summarize research results. Failed experiments need to be specifically pointed out, in order not to let others stumble on the same stone. Such a platform can be built through industry associations, such as American Association of Newspaper Editors (American Press Editorial Association), or online News Association (Online Journalism Association).

In many other industries, companies are reluctant to share their findings with others, or to co-create a platform or infrastructure. But the status and nature of newspaper media industry is a little special. There is little direct competition between news organizations. More importantly, as many journalists and news editors say: The pursuit of openness by journalists is in itself the embodiment of the value of the news industry.

Complete the full text.

Want to learn about website operations? I'll teach you! Under