How to redesign the mature feature


What do your customers want exactly

I’m a product manager, in charge of the web crawling product that can help users to crawl data from the web.

In the previous version, we found a tendency that more and more users complain about the inconvenient operation while creating a web crawling task.

So we used some tech to observe our users’ operation path, they showed many scenarios to describe how awful of their experiences.

For example, if a user wants to crawl data from the page, he firstly should select the first element, then create a temporary list to store it. Finally, he has to choose the rest of the elements to finishing this process;

The worse thing is that users would be lost in this long process because of the dozens of possible actions when creating a temporary list.

You see, the whole process is very annoying and lengthy, so we product team had summarized two core principles:

  1. Give the results that users want
  2. Do not make users confounded

How to redesign the mature feature

Give the results that users want

What are 「Give the results that users want」?

It’s easy to understand, because of the inertial thinking. In the past few years, we usually think that our users are web crawling enthusiasts, they like that we provided all settings on the screen and they can choose what they want;

Yes, it’s high efficiency, but in this observation, we surprisingly found that we have much more junior users, that want the simple and more natural way to crawl data.

According to our long-term user behaviors research, most of the users’ elements selections are similar on the webpage
For example, these three titles of Amazon goods are in the same area
Amazon goods

So what we need to do, is to help users to select all kinds of similar elements of popular websites easily.

This process is fascinating. However, I can’t disclose more details because it’s the business confidential
After many times attempt, we finally released our research results into our products:

The first improvement is 「Similar elements recommendation」; once users click any element, the app would automatically identify and pre-select the other similar elements.

Similar elements recommendation-before
Similar elements recommendation-after

The second improvement is 「Child elements recommendation 」, once users had selected an area that including so many child elements, the app would automatically identify and pre-select all these child elements.

Child elements recommendation-before
Child elements recommendation-after

With these two kinds of improvements, users can easily select vast amounts of elements.

Do not make users confounded

The next point is 「do not make users confounded」, as you can see, there’re so many kinds of web controls:

  • text input
  • button
  • dropdown select
  • image
  • link

When users click different elements, there may indicate different goal:

click text input; the user may want to input some keywords to search and extract the data in the search result page

click dropdown select; the user may want to loop through options in the dropdown and extract the data after the value of dropdown changed

In the past, the app would show all possible selections no matter what option the user selected.

So I decided to simplify this process for users, I have listed out all web controls and communicate with users about the possible goals when they click different web controls.

Finally, we finished.

In the new version, the app would only show the most relevant options based on the element which the user clicks.



After releasing the new version, some of our loyal users complain about this release.

We talked to these users and improved them again and again. Finally, most of the users accepted them.

We have some key indicators to guide us to validate our many improvements, after releasing the new version, the 「average success creating tasks rate」 is increased 30% and the 「average creating tasks time」 is decreased 20%, which prove that our improvements are successful:)

Of course, our new user’s retained rate is also increased a lot.


  1. You should figure out the variation of the composition of your primary users.
  2. In most of the times, redesign the mature features means reconstruct users’ daily habit, which means that you have to prepare as much as you can
  3. Before the redesign, it’s better to talk with different kinds of users such as loyal users, new users, and the mid users, try your best to describe all types of their operation path of this feature and abstract the common questions and solve them.
  4. Although you have done enough preparation, the old users would still complain the redesign results at the first beginning, do not worry about, relax and talk to them, you would get something that you might dismiss.
  5. Preparing the data track solution on each improvement, you would find something surprise that users would not say.


Leave a Reply

Your email address will not be published. Required fields are marked *