Causes of failure during the classic Project Process

After 20 years of application, people found that there are many problems in the process of classic projects, which forces people to study the existing problems.
To find a solution. The most difficult problem with classic project management is that the project is difficult to plan and the plan made is usually
Error. As a result, the development team often goes to two extremes: either not planning, such a group cannot answer "you guys
When can it be completed ?" This is a problem. Either invest a lot of energy in the plan until you are confident that the plan is positive.
True, but such "correctness" is often self-defeating. Such a plan may be more comprehensive, but it does not mean more accurate or more
. Because of the characteristics of the software, uncertainty always exists.
Although classic project management regards cost, quality (function goal), and time as the most important triangular constraint, it is a pity
Many researchers have told us that traditional methods may not necessarily produce satisfactory answers. The following are the conclusions of many researchers.
Result:
About 2/3 of projects significantly exceed the budget (Lederer and Prasad 1992 ).
64% of the features in the product are rare or never used (Johnson 2002 ).
Generally, the project spends more than 100% of the Schedule (stndish 2002 ).
This forces us to carefully investigate the causes of project failure, which can be summarized as five reasons.
1) The estimated deviation in the initial stage may be large.
It shows the range of uncertainty considered by Boehm in different aspects of the sequential development process. This figure is called inaccurate.
Qualitative taper. The figure shows that the estimated progress deviation may reach 60% to in the project definition phase ~ 160%, that is
The actual cost of a project completed in the pre-period of 20 weeks may be 12 ~ Within 32 weeks. After writing down the requirements, it is estimated that
The deviation may still be ± 15%. At this time, the expected 20 weeks may actually take 17 ~ 23 weeks. Such a large estimation bias
However, effective project management cannot be implemented.
2) Activities will not be completed in advance
Imagine a Senior Programmer is preparing an interesting new function for a product, and
How does cmme allocate time for reviewing and preparing documents? Without a doubt, he will spend most of his time compiling
This interesting program, and the preparation for review only leaves enough time, and it will not be finished until the end.
In fact, this behavior is so common that there is a name called Parkinson's Law (1993 ):
The task is always completed until the last moment ."
If a job is assigned five days on the Gantt chart, the employee who handles the activity will certainly spend five days, even if
He will also try to use some new hot technologies by adding some fancy modifiers (gold-plated requirements.
In many corporate cultures, if an activity is completed in advance, it is often accused of being inaccurate in his estimation.
Why should we take this risk to complete other tasks in advance? The human nature is like this: Use extra time
Do something that is valuable to you but not useful to others.
3) The delay is passed down along the schedule
Because traditional plans are activity-based, they focus primarily on dependencies between activities. Evaluate the knowledge of Gantt
Figure, which shows four activities and their dependencies.
If you need to complete the test in advance, it is a coincidence that some events are required:
The encoding of the middle layer is completed in advance, and it is affected by the completion time of the activity of adding data tables to the database.
.
Encode the user interface in advance.
Arrange the test personnel in advance.
The key is that, even in such a simple case, three things must happen before the test starts,
However, any of the following things may delay the test:
The end time of user interface encoding is delayed.
The time spent on the middle layer encoding exceeds the plan.
The middle layer encoding takes time to meet the requirements, but it is too late to add data tables to the database, resulting in a delay in testing
Trial time.
No testers have been arranged.
That is to say, a lot of things need to be completed before starting the test. One of the things is delayed, which may lead to overall delay.
Error. As a matter of fact, we have confirmed that the activities are rarely completed in advance, so most of the things that happen are delays.
The delay will be transmitted along the schedule, and subsequent projects will rarely be started in advance.
4) features not developed based on priorities
The third reason why traditional project management methods may not necessarily bring high-value products is that the plan is not in line with
Values of users or customers are prioritized. Many traditional plans actually assume that work is acceptable.
To complete, the order of work is mainly determined by dependencies. However, as the project end time approaches, the development team will
Some features are abandoned to keep up with the progress. Because some features are not developed in order of function priority
More important than delivered functions.
5) Ignoring uncertainty
The traditional planning method does not recognize the existence of uncertainty because of its 4th shortcomings. People assume that the initial demand analysis is acceptable.
To produce a complete and complete definition of the product. We assume that the user will not change throughout the planned coverage period.
Their ideas will not be more detailed, nor will they put forward new demands.
Similar to this, we also ignore the uncertainty of how to build products, which is not discovered by a certain technology in implementation.
It must be appropriate, but we have allocated it a definite period of time (two weeks). In fact, we cannot expect a start.
Determine all activities required in the project process, but we do not want to acknowledge this.
After analyzing the problems existing in the traditional project management method, we will not feel disappointed with so many projects.
Strange. The activity-based planning method disperses our attention to functions, and functions are the unit to measure the value of customers.
The participants of the project considered multi-task processing as a solution to the delay with good intentions, but the result caused more project delays.
Error. When the remaining time of the Schedule table is not enough, some features will inevitably be abandoned, because developers follow the best
Effective development methods to arrange the progress, so the abandoned function is not necessarily the lowest value for the user.
Ignoring the uncertainty of user requirements will lead to the completion of the task on time, but many important functions found later are not
There is a way to join, or even if it is added but it is not a project delay, or it is not appropriate to reduce the quality.
Based on these analyses, we must consider solutions to these problems.