I know a lot of friends, netizens, their biggest troubles now is not difficult to do, but the boss does not pay attention to the test. The boss does not pay attention to the test work, and does not pay attention to the testers in China's software companies, especially in small and medium-sized enterprises. Why is this phenomenon, is it really not important? of course not! Industry insiders know that testing is an important means of ensuring software quality. I want to cause this phenomenon or "interest", because testing can't produce intuitive economic benefits, and the ultimate goal of each boss is of course benefit, so this will not be booked. But you can't expect to take a bunch of scrap copper rotten iron to buy a good price? So I think if the tester can reduce the cost while improving product quality, how do you open you?
If you want to do this, we have to solve two problems: 1. How to make your boss pay attention to testing; 2, how to reduce test costs.
Let's solve the first question first: How to make your boss pay attention to test. For this issue, many people have tried the truth, but it is difficult to produce good results. Now we may wish to change a way, speak with data, because the data is intuitive, more convincing.
In order to get a set of data, I made a simple to compute: assuming that there is a project group to develop products to customers, customers have a total of ten needs, regardless of size, I have assigned ten points to each requirement, then the entire product has 10 × 10 = 100 scores. Now let's take a look at how many scores left in the event of no test:
activity
Defect description (conservative)
Lose score
Remaining score
Demand collection
Customers have ten needs, but because customers do not fully know their needs or other reasons, the demand staff collects two needs when they collect them.
2 × 10 = 20
100-20 = 80
demand analysis
When analyzing 8 demand for collected, because each person understands different reasons, there are two demand and customer deviation of 50%.
20 × 50% = 10
80-10 = 70
Summary design
Because the designer's understanding of the needs is deviated, it may cause 50% of the functional error of a demand; at least 10% of defects may be generated due to technical issues during design.
10 × 50% = 5
70 × 10% = 7
70-5-7 = 58
detailed design
Detailed design defect 10%
58 × 10% = 5.8
58-5.8 = 52.2
coding
20% encoding defect
52.2 × 20% = 10.44
52.2-10.44 = 41.76
other
5%
41.76 × 5% = 2
41.76-2 = 39.76
According to this data, it can be seen that the demand for final software is approximately 39.76% of the original needs of customers. I would like to see this data. I should be a touch of the boss: Give such a product to the customer, can you still have a satisfactory return? However, the above data is based on my personal experience. Because of the different production capacity, this result is not the same, and each company can estimate according to its own actual situation, and have no testing capacity without testing. I think most of the final results of the company should not exceed 80%. If so, then the test is indispensable.
Speaking here, in fact, I have answered the question of this article title: It is to take a set of data belonging to you, put it to your boss, I believe there will be a sense of persuasion. Through these data, I still want to analyze my second question: How to reduce the cost of the test? Continue to analyze our data, resulting in the defect score generated throughout the development process: 100-39.76 = 60.24. Based on this, the defects generated by each stage account for the ratio of all defects:
stage
Defective value
proportion
Demand phase
20 10 = 30
30 / 60.24 = 50%
design
5 7 5.8 = 17.8
17.8 / 60.24 = 30%
coding
10.44
10.44 / 60.24 = 17%
other
2
2 / 60.24 = 3%
And the authority data I got in a book is:
Development steps
Defect ratio (%)
Preparation Demand Specifications
55
design
25
coding
15
other
5
The two groups of data are similar, whereby most of the defects are generated during the demand phase. If repair cost and release time are not considered, most of these defects can be discovered in the encoded test phase and the maintenance phase after the product release. However, the repair defect is cost, the research results of the US Quality Assurance Research Institute show that the soon the problem, the problem in the software is discovered, the lower the development cost; the cost of modifying software defects after encoding is 10 times before the encoding, in the product The cost of modifying the software defect after delivery is 10 times before delivery; the higher the software quality, the lower the maintenance cost after the software release.
Comprehensive deficiencies and repair cost ratios, we can get such a table:
stage
Defect ratio (%)
Modify defect price (conservative)
demand
55
1
design
25
5
coding
15
20
other
5
10
Maintenance phase
50-100
It is not difficult to see that the sooner we test, the easier it is, the more quality products, and the repair cost will be greatly reduced.
Therefore, we should fully understand the customer needs on the basis of the testers to fully understand the needs. Let test staff participate in the collection, directly and customer communication, and participate in the review and system testing of various instructions, and do a comprehensive control of software quality.
This can reduce the cost of the test, but also make it large enough to satisfy the customer's needs. I think this is also the result of each boss to see.
