How to improve the registration conversion rate by up to 12% through design?

I first thought that design was how to make a variety of novel graphics, textures and interfaces, chasing trends and ideas. But later found that the most difficult design is to balance the various factors, in the constraints of the rules to find a solution to go ahead and be seen value. The former is easy to meet, and the latter to do is very difficult, PM does not give force, communication is not smooth, development does not support, the boss is not satisfied. Many designers will worry about the cause and the solution is what, and just I recently in the circle of communication found a very serious phenomenon: first-line designers for data and target sensitivity is very low, so the design is not convincing, since the design is very good things others can not understand, pushing very difficult. And many product managers also think that designers are stubborn artists, do things in addition to beautify see no specific value.

So why does understanding data help to improve these design problems?

1, the Product manager, the company leaders all value the data, the big goal usually is not the user growth data, the product active data is the commercial data. So the data is the real concern of the partners and attention to the problem, is their KPI, often product managers to give a lot of designers need is to improve a data index;

2, the data is objective, the design plan after the on-line through the correct analysis can see good and bad, the value of the design can be very intuitive performance and be measured;

3, the goal of the data can be translated into the abstract concept of "language" (at least the PM and research and development are understood), for example, you said, "This program can make the list more simple and clear, easy to input, the success rate is higher than 5%" than that "this program can make users experience better", the former is more convincing;

4, the data is not only the product manager's job, the product manager also sees but they are more to pull out the influence of the demand aspect. Product Manager in the design is not professional, if the other side of the interpretation of course is not professional advice, designers need to dig their own data and figure out how to use design to solve the essence of the problem;

Directly throws our team of data using a set of design methods to "identify data targets – dig deep data – fast Agile validation design – ensure true and efficient":

　　I. Determination of data targets

Some time ago our product director to find my slot registration process design is too bad, so I must make a good change. But he is nothing but the text is not beautiful, the button is too small, the map is not good-looking and so very abstract concept. So my heart Orz is "Why use your own preferences to measure the design is good!", then pulled up the product manager to check the registration conversion rate. Our registration conversion rate is 70%, but is this effect good or bad? If I want to optimize, what is the goal of my optimization?

So before you do design, be sure to determine the current data situation, and set objective measurable data optimization objectives:

1, by comparing the competition: for example, we are social products, through the convenience of large companies to find some and our similar process of competition, found that their registered conversion rate on average 75%, excellent products can reach 85%.

2, the use of industry data and communication: if there is no channel to understand the competition can be found on the Internet or search information, and can also ask friends in other companies, former colleagues or predecessors, although the data is more sensitive but they generally can give the general advice.

Finally, the product director and I said to optimize the design can, we probably set the target to achieve registration conversion rate of 80%, during which I try several methods, you do not care whether you feel like, take the results to speak.

Second, deep digging data and analysis, to find design problems

Get the design goals, the designer immediately to the entire process of experience analysis, found a lot of possible destructive experience problems, but in fact this is wrong. The designer's evaluation is also out of the subjective experience of the designer, take their own understanding of the user to say a blind guess, the design of something persuasive must be very low, porous. So before you do, you should first peel the data one layer at a time, dig deep into details, and locate the key points:

First found in the registration three steps, step one and step two of the loss rate is very large, is needed to focus on the place. such as the loss rate of the registration step two, we define the user clicks the mailbox registration, but did not click to complete:

1 Enter and exit without any input.

2 after input because of various input problems, format errors, existing problems caused by registration can not be completed

3 network problem, service side problem leads to not really complete the registration of account

How do you know who is the mastermind? The cheetah PM is usually used to bury the elements before the product is on line, so we just need to check every element's buried point data. Finally found that one of the "Account and password format error" Pop-up ratio is high, accounting for the "Next" button PV 10%, so the suspicion is that users often enter the wrong format to lose. With the data, the designer immediately know where the problem is, "before for the simplicity of the interface, the format of the wrong window is the user input after the completion of the click will be detected," after the modified hint exposed, and the correct input button will be lit. This positioning and optimization of the step after the loss of the big head is immediately resolved, and PM also think the designer is very sharp.

　Third, quick and agile Verification design plan, get the data conclusion as soon as possible

Another example, step one without any click behavior on the loss of the user reached 14%, the user downloaded the app and opened the first page without doing anything to go, very strange:

1 Users find it necessary to register, too troublesome to try (but we can not remove the registration process)

2 After the user downloads, only the elements seen on this page are not enough to attract him

Because it involves the user's subjective feelings, the problem is not so clear, the design method is to put all the elements of this interface, and then the brain burst a lot of optimization ideas, pick some good ideas for quick verification:

1) Modify the background video, try the European and American Beauty, local beauty, some people communicate with the scene

2 Add a new carousel map to tell users what the function and content of products, expect users to change interest

3 optimize the layout of the button and copy

The process of agile verification is to try this design first with the simplest scheme, and then two days on line to observe the data. Two days from now, if the data is going up, then deepen it in this design direction. If the data does not rise or become worse, remove, and then another design idea on line.

Soon we found the effect, such as our background video before a few people communicate the scene, we changed to the local beauty (Asian faces), found that the user clicks the button to raise a point. Then we tried the next version of the European and American Beauty, the data raised a point (it seems that the European and American Sister taller).

And the improvement button is more interesting, first of all, we suspect that overseas users worry about privacy, the Facebook button next to the privacy assurance of the copy, the data up a point. Changing the Sign up by mail to Sign up to free can increase by 3%, and even the "use of Facebook" button color and Facebook is consistent with the promotion of transformation.

After our series of efforts, the loss rate of the first step is reduced to 8%, the overall registration conversion rate has risen to 82%, the leadership is very satisfied and in the large-scale promotion of saving a lot of "expenses."

But not all design is a breeze shun, for example, in the verification process we learn a lot of apps do some of the wheel map introduction product highlights, but after a few days found that the loss has not decreased but also slightly increased. So also do not have to work hard for several days of the beauty map, had to go offline.

Visual designer also very surprised, a pair of "how can?" I think it's pretty good. "Expression, and then we summarized the reflection is because all users are downloaded from the GooglePlay app, and the store screenshot has been very good description of the product highlights, the new addition of the Carousel map not only repeat, but also make the page become complex , the user is overwhelmed by the operation.

　Iv. ensure the authenticity and efficiency of the verification process

1, the same version verifies only one design point

At that time we modify the button copy when the development of the background video can also be conveniently replaced, the cost is very low. But if the same is a step of the optimization if the same time on the line, we do not know who has brought the effect, so deliberately send a version of the button, the background video replacement on the next version of the online. But this requires very fast version iterations, and the cheetah's agile culture is publicly appreciated, so we can basically make a version every 2 days during the optimization.

2, rough first, then meticulous

Isn't it crazy to be a designer so fast? Where do you come from?

So what we often emphasize when we validate new ideas is "go first," casually online pick up an icon to do an interface, or preferably do not need to design, you can use the lowest cost of existing elements to spell out the best, if this idea does not work, we will cut off after 2 days, if there is a positive role, such as 1%, The designer can start to seriously make it into 5%.

3, keep the data stable.

Maybe a lot of people are thinking about this 1%, 2% of the ups and downs how to determine is and design improvement related? The first thing to ensure that the data has a certain level, such as daily installation to 5K, the loss of 1% is 50 people, is a phenomenon that can explain the problem. At that time our products have been to maintain a certain amount of promotion, and not just to register this thing, but if there is no certain user and data to support, all about the quality of product improvement is bullying, do not own behind closed doors.

There is the stability of the data, such as we in the optimization of the registration is in so many places to advertise, if the advertising position of a change to attract the installation of the user changes, from the sister suddenly turned into an aunt, then our verification results are not credible, so do not do the design of their own things, this period and the promotion of the Department's cooperation is very deep.

Finally, because some of the product details are relatively sensitive, so I put the "registration conversion rate" case of the data and processes are blurred, only for people to understand more simple examples. It's a very complicated thing to do and it takes a lot of time and effort, but if you have the idea to be ready to explore, you want to "identify data goals – dig deep data – fast Agile validation design – and make sure that it's real and efficient" this approach can give you some ideas.