当前位置:网站首页>[Fantan] how to design a test platform?
[Fantan] how to design a test platform?
2022-07-07 17:07:00 【I'm going to have a hot meal】
Write a test platform , What's the most important ?
Code level ? Important but not the most important , After all, software with high code quality but difficult to use is everywhere .
Algorithm speed ? Execution requires 0.3 The second and 0.1 second , Colleagues in the company actually don't care too much .
Technology stack ? Nothing to say , Use mainstream and unpopular , Users don't care .
High quality bug Less ? This is the necessary foundation to become a usable platform , If bug many , That's not a good thing , You can't live at all .
Look beautiful ? Not to mention that they all use the same front-end style framework, which is not bad . Just compare the internal platform of the company , Who will say that the platform is excellent because of its beautiful appearance ? Especially leadership , Instead, you will often blame you for spending too much energy on the appearance, resulting in weak background functions .
So what exactly is the key to writing a good test platform ? After you read the following two examples of other industries , It is estimated that you will have your own answer ...
There is an old brand of cookies in the United States , More than 100 years of history , The cookies are very delicious , Challenge its opponents one after another , But it all ended in failure .
But this old brand was last year , Be completely terminated by an emerging brand , Just two or three years , Will be exceeded several times ...
Why can this emerging brand beat the old trump of biscuits so easily ? The reason is that emerging brands are not made of biscuits , It's the self-supporting biscuits of a large supermarket chain .
original , This supermarket has tens of millions of members , Receive hundreds of thousands of requests and feedback about cookies every day . What cookies sell best , What kind of biscuit is hard , What biscuit is soft .... And so on / Suggest / Make complaints / feedback All are sorted and counted by the supermarket , Finally, they follow the needs of the vast majority of customers , Made a proper biscuit .
Once on the market , Quickly packed shelves and shares , Rave reviews . And they even continue to optimize after starting so high , Dynamically adjust the ingredients and price of biscuits according to the real-time super large data feedback . This so-called biscuit finally dominated the whole American market ... And it's almost impossible for other brands to turn the tide ...
Another short story : Say there is a chain hotel , Many tenants complain about pillows every day , Some say pillows are hard , Some say pillows rebound slowly , Some say pillows are low , Some say pillows are airtight ....
In the long term , Finally, the hotel decided based on the huge feedback data , Make a pillow that suits the most people , After putting into use , The number of complaints drops instantly , Rave reviews .
In the above two stories , What is the secret of success ?
Obviously , yes data .
It's huge feedback and specific proportion .
If say : You know some people complain that pillows are long , Some people complain that pillows are short , Then you give yourself a reason : All tastes . Then don't optimize the change .
And if there is specific data support , You'll find out , Complain about the length , Complaints are short 99%, Complaints are only 1% .
Now , You just need to meet the interests of the vast majority of users , Lengthen the pillow to make it successful . Or prepare for the hotel 1% About a number of shorter pillows will be more considerate if they are left to a very small number of customers who complain about being long .
Without these specific figures , Then it will become a headless fly , Shut oneself up in a room making a cart .
The topic goes back to testing platform :
In addition to the factors mentioned at the beginning of the article , What's the most important ? It's design .... And the design is to be born on the basis of huge data .
You need a lot of feedback , Many suggestions , A lot of experience can be learned , Even a lot of complaints and roast . In order to really do a good test platform .
What function is used by fewer people , What functions are troublesome , What features like the most , What functions are urgently needed .
Only by understanding these , The design will be excellent , This is why I will give priority to explaining the data embedding points and statistics in the training .
(6 This month, the training homepage is designed )
Instead of making decisions in the company , Then the hard-working things were hidden by snow .
Bloggers many years ago , I have suffered such losses several times , At a price , After a painful experience , Just understand this truth . As a reader, you only need to spend 5 After reading this article in minutes, you will understand , Make a .
Of course, after understanding this truth , I immediately took measures , That is to completely open source the source code of various test platforms before , Even hand-in-hand 0 Basic course .
Publish these to the technology community , Let all peers comment , Well intentioned , Malicious , Something nice , To make complaints about Tucao , Take all the orders .
Then I have precious data , Big data .
For example, the interface test platform series of this official account :
Look at the number of visitors in a few chapters :( Absolutely true data )
Data factory series :
In thousands of uses , Catch up with hundreds of discussions every day , The platform is constantly optimized , Constant iteration .
This is the way to excellence .
But the beauty is , Once the article of official account is sent , You can't modify , It can't be revoked . So on the bumpy way forward , The feedback of those precious super large data , It can only be used in future chapters .
If you are anxious to learn , You can pay attention to my latest training , The interface test platform will use all these valuable empirical data , To reconstruct and become the current optimal level .
but , Iteration is endless , No best , Only better .
Official account interface test platform version , Future functions will also absorb these excellent design suggestions , To satisfy all readers and fans .
边栏推荐
- skimage学习(1)
- 智慧物流平台:让海外仓更聪明
- 直接上干货,100%好评
- 99% 用户在 Power BI 云端报表常犯错误
- Inner monologue of accidental promotion
- centos7安装mysql笔记
- LeetCode 1696. 跳跃游戏 VI 每日一题
- Talk about the realization of authority control and transaction record function of SAP system
- Shallow understanding Net core routing
- LeetCode 1981. Minimize the difference between the target value and the selected element one question per day
猜你喜欢
Pisa-Proxy SQL 解析之 Lex & Yacc
node:504报错
如何在博客中添加Aplayer音乐播放器
掌握这套精编Android高级面试题解析,oppoAndroid面试题
How to add aplayer music player in blog
skimage学习(3)——使灰度滤镜适应 RGB 图像、免疫组化染色分离颜色、过滤区域最大值
[image sensor] correlated double sampling CDs
Master this set of refined Android advanced interview questions analysis, oppoandroid interview questions
Master this promotion path and share interview materials
Sator推出Web3游戏“Satorspace” ,并上线Huobi
随机推荐
Skimage learning (3) -- gamma and log contrast adjustment, histogram equalization, coloring gray images
LeetCode 312. 戳气球 每日一题
DNS 系列(一):为什么更新了 DNS 记录不生效?
[image sensor] correlated double sampling CDs
《产品经理必读:五种经典的创新思维模型》的读后感
LeetCode 152. Product maximum subarray daily question
LeetCode 1049. Weight of the last stone II daily question
LeetCode 1774. The dessert cost closest to the target price is one question per day
As an Android Developer programmer, Android advanced interview
QT 图片背景色像素处理法
[designmode] proxy pattern
LeetCode 1981. Minimize the difference between the target value and the selected element one question per day
LeetCode 1031. Maximum sum of two non overlapping subarrays
Skimage learning (1)
rpcms获取指定分类下的文章的方法
Lie cow count (spring daily question 53)
dapp丨defi丨nft丨lp单双币流动性挖矿系统开发详细说明及源码
一文读懂数仓中的pg_stat
Test case management tool recommendation
[Seaborn] implementation of combined charts and multi subgraphs