Sources of Requirements
Good requirements start with good sources. Finding those quality sources is an important task and, fortunately, one that takes few resources. Examples of sources of requirements include:
- Administrators and maintenance staff
- Domain Experts
- Industry Analysts
- Information about competitors
Requirements Gathering Techniques
After you have identified these sources, there are a number of techniques that may be used to gather requirements. The following will describe the various techniques, followed by a brief discussion of when to use each technique.
To get the requirements down on paper, you can to do one or more of the following:
- Conduct a brainstorming session
- Interview users
- Send questionnaires
- Work in the target environment
- Study analogous systems
- Examine suggestions and problem reports
- Talk to support teams
- Study improvements made by users
- Look at unintended uses
- Conduct workshops
- Demonstrate prototypes to stakeholders
The best idea is to get the requirements down quickly and then to encourage the users to correct and improve them. Put in those corrections, and repeat the cycle. Do it now, keep it small, and correct it at once. Start off with the best structure you can devise, but expect to keep on correcting it throughout the process. Success tips: Do it now, keep it small, and correct it immediately.
Conduct a brainstorming session
Brainstorming is a short group session where everyone is allowed to say whatever they feel is important to the topic of discussion. After that, a facilitator leads the group in organizing and prioritizing the results. The following basic rules for brainstorming ensures better results:
- Start out by clearly stating the objective of the brainstorming session.
- Generate as may ideas as possible.
- Let your imagination soar.
- Do not allow criticism or debate while you are gathering information.
- Once information is gathered, reshape and combine ideas.
Face-to-face contact with users through individual interviewing is the primary source of requirements and an important way you gather and validate their requirements. Remember that it is not the only possible technique, and that you can conduct interviews many different ways. Develop a repertoire of styles to fit different situations. Unless you use the system yourself, you will need to make an effort to understand and experience the user's problem to describe it clearly and correctly.
If face-to-face meetings are possible, they are always preferable, because they provide a better means of uncovering the problem behind the problem. Sometimes, though, face-to-face meetings with stakeholders are not feasible (when developing products for the consumer market, for example). In those situations, consider using questionnaires. Send a set of questions, possibly with multiple choice responses, to the relevant stakeholders, and ask them to complete it and return it to you. Success tips: Keep it short and given them a deadline.
This technique has the advantage of providing a lot of information for statistical analysis. However, the questions must be well designed to be clear and to avoid so-called "leading questions", which bias the responses.
Work in the target environment
Experience the work of the users for yourself. Working with users helps you understand problems that have resisted previous solutions. Familiar systems developed in this way inevitably include tools for programmers, such as interactive editors and compilers, as the developers naturally have both the expertise in the subject area, and the desire to solve their own problems. It would be good to see the same dedication devoted to solving problems in other areas too. Where the work cannot easily be experienced in this way, it may still be possible to do a bit more than just sit quietly and observe. Users can give you a commentary on what they are doing, what the problems are, and what they would like to have to make the work easier.
Study analogous systems
The starting point for many projects is often a similar or an existing system. Sometimes, comparable products and systems contain working versions of good ideas for solving user problems. You can save the time lost in reinventing the wheel by looking at systems already on the market, whether they are systems installed at the user's site or products made by rival organizations. Even if they are trying to solve slightly different problems, they often provide valuable clues as to what you need to do.
Listen when a customer asks why a product couldn't do something that the customer wants, and keep a list of these suggestions. Later, use it to start discussions with other users. You should be able to obtain some requirements directly this way. If not, capture and store suggestions for future use.
You can describe to users selected features of other products. Explain that the system is designed for another purpose but contains an interesting feature, and you wonder it or something similar would help them. Sometimes these systems are described in documents, such as a contract from another organization or a report written for management. Often, these documents were never intended as formal requirements, and were written merely to communicate a stream-of-consciousness idea. Define a process of going from disorganized to organized information.
Such a process might involve the following activities:
- Read the document from end to end (several times) to comprehend what the customer wants and what actually has been written.
- Classify all of the types of information in the document. (user, system requirements, design elements, plans, background material, irrelevant detail)
- Mark up the original text to separate out such requirements.
- Find a good structure for each of the different types of information such as: a scenario for the user requirements, functional breakdown for the system requirements, and architecture for the design.
- Organize the information to show gaps and overlaps. Feel free to add missing elements, but confirm these decisions with the users.
- Create traceability links between these information elements to show the designers exactly what the users want.
- Convince the customer to accept the new information as the basis for the contract.
Examine suggestions and problem reports
Requirements can come from change suggestions and user problems. A direct road to finding requirements is to look at suggestions and problems as first described. Most organizations have a form for reporting system problems or software defects. You can ask to look through the reports (and there will probably be many). Sort them into groups so you can identify the key areas that are troubling users. Ask users some questions about these areas to clarify the users' actual needs.
Talk to support teams
Most large sales organizations have a help desk that keeps a log of problems and fixes, and support engineers who do the fixing. Many organizations have similar facilities to support their own operations. Talking to the help desk staff and the support engineers may give you good leads into the requirements, and save you time. Also talk to the training team and installation teams about what users find to be difficult.
Study improvements made by users
This is an excellent source of requirements. Users of a standard company spreadsheet may have added a few fields, or related different sheets together, or drawn a graph, that exactly meets their individual needs. You need only ask: Why did you add that? Their answers help you get to the heart of the actual requirement. This applies also to hardware and non-computer devices. For example, a lathe operator may have manufactured a special clamp, or an arm that prevents movement of the tool beyond a certain point. Any such modification points to something wrong with the existing product, which makes it a valid requirement for the new version.
Look at unintended uses
People often use things for purposes for which they were not designed. This is a good way to get new ideas and to think of innovations. For example, an observant product manager noticed that an engineer was staying in the office late to use an advanced computer-aided design system to design a new kitchen layout for his home. Inexpensive commercial products are now widely available for home use.
Workshops can rapidly pull together a good set of requirements. In two to five days, you can create a set of requirements, and then review and improve them. If everyone in a workshop tries to estimate the cost and value of each requirement, the document becomes much more useful and cost-effective.
Workshops are quicker and better at discovering requirements than other techniques, such as sending questionnaires. You are bringing the right collection of people together, and getting them to correct and improve on their requirements document.
A workshop is inherently expensive because of the number of people involved, but it saves a large amount of time. If you can define the product right the first time and cut three months off the requirements gathering, the savings could be enormous. The workshop has to be thoroughly organized to take advantage of people's time.
Choose a quiet location for the workshop so that people are not disturbed by day-to-day business. Mobile phones should be discouraged; arrange to take messages externally. Take advantage of informal interactions by choosing a site so that people don't go home at night or go out separately.
Demonstrate prototypes to stakeholders
Prototypes allow us to immediately see some aspects of the system. Showing users a simple prototype can provoke them into giving good requirements information or changing their mind about existing requirements. The techniques described here help you gather ideas for requirements. Prototypes and models are an excellent way of presenting ideas to users. They can illustrate how an approach might work, or give users a glimpse of what they might be able to do. More requirements are likely to emerge when users see what you are suggesting.
A presentation can use a sequence of slides, storyboard, an artist's impression, or even an animation to give users a vision of the possibilities. When prototyping software, make a mock-up of the user interface screens, emphasizing that there is no code and that the system has not been designed or even specified yet (fair warning: there are dangers here for the unwary).
This prototyping aims to get users to express (missing) requirements. You are not trying to sell users an idea or product, you are finding out what they actually want. Seeing a prototype, which invariably is wrong in some ways and right in others, is a powerful stimulus to users to start saying what they want. They may point out plenty of problems with the prototype! This is excellent, because each problem leads to a new requirement.
Which Technique to Apply?
Which technique to apply depends on a number of factors, such as:
- Availability and location of stakeholders
- Development team knowledge of the problem domain
- Customers' and users' knowledge of the problem domain
- Customers' and users' knowledge of the development process and methods
Characteristics of good requirements
The characteristics of good requirements are variously stated by different writers, with each writer generally emphasizing the characteristics most appropriate to their general discussion or the specific technology domain being addressed. However, the following characteristics are generally acknowledged.
|Unitary (Cohesive)||The requirement addresses one and only one thing.|
|Complete||The requirement is fully stated in one place with no missing information.|
|Consistent||The requirement does not contradict any other requirement and is fully consistent with all authoritative external documentation.|
|Non-Conjugated (Atomic)||The requirement is atomic, i.e., it does not contain conjunctions. E.g., "The postal code field must validate American and Canadian postal codes" should be written as two separate requirements: (1) "The postal code field must validate American postal codes" and (2) "The postal code field must validate Canadian postal codes".|
|Traceable||The requirement meets all or part of a business need as stated by stakeholders and authoritatively documented.|
|Current||The requirement has not been made obsolete by the passage of time.|
|Feasible||The requirement can be implemented within the constraints of the project.|
|Unambiguous||The requirement is concisely stated without recourse to technical jargon, acronyms (unless defined elsewhere in the Requirements document), or other esoteric verbiage. It expresses objective facts, not subjective opinions. It is subject to one and only one interpretation. Vague subjects, adjectives, prepositions, verbs and subjective phrases are avoided. Negative statements and compound statements are avoided.|
|Specify Importance||Many requirements represent a stakeholder-defined characteristic the absence of which will result in a major or even fatal deficiency. Others represent features that may be implemented if time and budget permits. The requirement must specify a level of importance.|
|Verifiable||The implementation of the requirement can be determined through basic possible methods: inspection, demonstration, test (instrumented) or analysis (to include validated modeling & simulation).|
There are many more attributes to consider that contribute to the quality of requirements. If requirements are subject to rules of data integrity (for example) then accuracy/correctness and validity/authorization are also worthy attributes. Traceability confirms that the requirement set satisfies the need (no more - and no less than what is required).
To the above some add Externally Observable, that is, the requirement specifies a characteristic of the product that is externally observable or experienced by the user. Such advocates argue that requirements that specify internal architecture, design, implementation, or testing decisions are probably constraints, and should be clearly articulated in the Constraints section of the Requirements document. The contrasting view is that this perspective fails on two points. First, the perspective does not recognize that the user experience may be supported by requirements not perceivable by the user. For example, a requirement to present geocoded information to the user may be supported by a requirement for an interface with an external third party business partner. The interface will be imperceptible to the user, though the presentation of information obtained through the interface certainly would not. Second, a constraint limits design alternatives, whereas a requirement specifies design characteristics. To continue the example, a requirement selecting a web service interface is different from a constraint limiting design alternatives to methods compatible with a Single Sign-On architecture.