When AI was involved in the design of the facelift, their new product development efficiency increased by 80% →
2025-02-21
Zhiyi Technology
Sister Amy isThe design director of a Jiangsu, Zhejiang, Shanghai and Tao brand, and also a long-time user of [Zhiqian], she leads a team of nearly 20 designers, operating two major Korean casual women's clothing brands with annual sales of more than 100 million. Since its inception in 2015, the teamThe monthly GMV is stable at 20 million 。
A while ago, Zhiyi visited Sister Amy's company. In this issue, we will compile the content of the return visit into a document.Let's take a look at what changes have been made to the opening work of e-commerce designers with the help of AI big data after more than two years of "knowing money".？
![]()
How to open a payment for an e-commerce clothing company
In order to ensure sales and control the cost of trial and error, when Amy's team develops new models every quarter, they usually follow the best-selling version that has been verified by the market, and then iterate with the popular elements of the season. Sister Amy revealed that nearly ninety percent of the company's new models are so remodeled, and they will also do comparative analysis with Internet celebrity bloggers and similar brands.
However, this seemingly safe method cannot avoid two major pain points in practice:
1) Finding the elements is like looking for a needle in a haystack
Designers need to find popular elements on platforms such as Xiaohongshu and TaobaoanalyseThe style diagram refines the key points.ThisManual sieveSelection is time-consuming, each identified epidemic point to average to4-5 days, even if I have a number of favorites100 articlesOutfitsnotemostAfter you can only get one or two useful information。
2) What the platform pushes is not necessarily really popular
Xiaohongshu's recommendation mechanism can easily form an "information cocoon", which makes it possible for the team to obtain information that is out of touch with real market trends, and it is difficult to find styles and outfits with corresponding elements.
In addition, more than 20 designers have their own opinions on the company's design tonality, and when the meeting is held, there are often disagreements within the team.For example, some people think that the bow is hot, and some people think that metal buckles are a trend.
In Sister Amy's opinion,These cumbersome hidden costs "eat" the time and energy of designers in the accumulated work.If a misjudged pop element makes it into production, it is likely to lead toMillion-level inventory backlog has become an "out-of-season model" that can't be sold。
In order to give us a more intuitive understanding,Sister Amy demonstrated their problems and solutions to us through two specific work scenarios。
![]()
Sister Amy's Problems and Solutions
Scenario 1: How to find the right elements for the facelift?
Problem: When we visited, Sister Amy's team was planning the development of a new coat, because the Gray crewneck coat launched the previous year sold well, and many customers reported that the version was very good, so they wanted to integrate the popular elements of the current domestic market on the basis of retaining the housekeeping version.
According to the past facelift method, even if the elements are added or subtracted from the existing version, it will take five or six days for the AMY sister company to modify the proofing, and the designer has to rely on experience to guess blindly: the elements may be piled up too much, the style may be deviated, and the sample clothes that take a week to get are likely to be rebuilt.
Solution:Use [Zhifu] to lock the current popular elements, and use data (likes/comments/favorites, etc.) to evaluate the performance of the elements。
Through [knowing].【Data analysis】Function, after the Amy sister team intelligently screened according to the style and tonality of the brand, you can observe the recent data performance of the "Korean casual" style coat in Xiaohongshu, click【View Style】You can also directly browse the details of related outfit notes.
Source:KnowProduct interface
On the Related Notes page under the [Korean Casual] tab, select[Released in the past 30 days], [Most liked]sorting, from which you can noticeThe main design element of the Korean casual coat that has become popular recently is the stitching of fur or khaki, which is more consistent with the trend of the popular Bath style in the autumn and winter of 24, which emphasizes the collar to be made of leather, lamb's wool, and suede materials。
Source:KnowProduct interface
In order to focus more on the design details of the collar, Sister Amy followedOriginal versioncoatsCrewneck, follow the frontThe operation narrows down the benchmarking and is used to determine the "crewneck coat"More specific design element trends。
You can notice it on the note page associated with the [Round Collar].In the new season of crewneck design, the notes with good data performance are mainly toPolar fleece/fur splicing process is the mainstayThis finding can also be used as an important reference.
![]()
Source:KnowProduct interface
After querying the high-profile notes and style pictures of [Korean Casual] and [Round Collar] in Xiaohongshu in turn, with intuitive data as support, the designer has a rough idea of which elements are more popular in the near future.
If further market research is required on the target design element, it can be passed next【Smart Search】Box select the element, and you're doneOne-click direct access to e-commerce platforms such as Taobao, Douyin, and Dewu, and find style drawings and product links with similar elements。
Source: Intelligent Image Searchinterface
Then sort [Shelf Time] and [Sales] as needed, and you can also use data to compare the high-sales and popular models on each platform to verify whether the design element has room for sale.
Scenario 2: How to make the design elements more in line with the brand tonality?
Challenge: After the design elements have been preliminarily determined,Sister Amy will also double-verify - according to the companyThe brand style tonality is excavated in the same type of bloggers that we have paid attention to before, so as to form an accurate facelift reference.
In addition to finding popular elements on social media such as Xiaohongshu, some brands also have their own "blogger library", which collects a group of online bloggers whose style matches the brand's tonality.
But in this case, Sister Amy thinks that the efficiency of finding elements is relatively low, and what is more troublesome is that it is impossible to effectively verify whether the blogger's style is in line with the current fashion trend.
Solution:Use the [Subscription] and [Smart Search] functions of [Zhifu] to determine the matching degree between the element and the style of the store, and dig and purchase this type of style for dismantling reference across the network.
In the knowing【Subscription】module selection coat category,Sister Amy can be on her own【Main store benchmarking】Group Selection【The most likes in the past 30 days], can be observed to be distinctive, and easier to change is the splicing fur, khaki or fleece collar, the result is consistent with the previous one, indicating that the previous idea of finding elements is correct.
Source:KnowProduct interface
In the meantime, if Sister Amy finds a good reference style, pass【Smart Search】You can also mine the data performance of the target model on Taobao, Xiaohongshu, INS and other platforms.
Source: Intelligent Image Search界面
包含近30天内【销量最多】、【发帖数最多】等数据维度都支持按需排序，再通过数据的比对，评估看中的设计要素是否真的在市场比较流行，最后再采买或与供应链沟通。
That's it for a review of this issue of the user story. By withAmy姐及设计团队的谈话，我们深刻了解到，从两年前开始使用多【知款】以来，他们With the help of AI artificial intelligence and big data analysis, the company's quarterly development of new models has also been more controllable risk avoidance。
接下来，我们还将根据最新收集到的客户需求进行复盘，不断开发出更多实用的AI功能、产品，敬请期待~
👇🏻 往期服务案例 👇🏻
![]()
