Discussion Locked

3 Important issues

0 helpful
15 replies
Closed to new repliesThis discussion is closed. It stays readable, but it takes no new replies.

What are your viewes on the following which I beleive are very important issues in project management?
1. What is the best way to obtain user requirements?
2. What are the most common reasons for the system not meeting the user requirements?
3. How do you handle user who constantly change their mind about what they want?

15 replies

by softcorp.us

Sounds like you are having the traditional universal IT project plan problems with users/clients.

The abbreviated answers are:

Q1. What is the best way to obtain user requirements?
A1: Meet face-to-face, discuss, write a document that includes all details and schedule milestones, and get formal sign-off.

Q2. What are the most common reasons for the system not meeting the user requirements?
A2: User and/or analyst did a poor job of defining them.

Q3. How do you handle user who constantly change their mind about what they want?
A3: Meet again, redefine the requirements, rewrite the document, and get formal sign-off again, and try to obtain schedule adjustments.

Users changing their mind or making requirements changes has been part of IT projects since there were IT projects. It will "always" happen so just expect it. The key to success is documentation and communication. Act pro-actively with the user at all times. The project plan should clearly establish when requirements changes are no longer permitted in order for the schedule to be met. These may be violated, but establishing them helps create a sense of urgency and responsibility.

You will often find that requirements changes will be made without allowing any additional time. So, the formal documentation and sign-off will help explain clearly why there are delays, why the schedule is tight, etc. If this is done professionally, it won't seem to be as much of a CYA matter. Requirements changes are simply a part of almost any technical project and you have to roll with them (formally and professionally).

by j.lupo

My doctorate deals with this exact issue. My dissertation is over 600 pages on this topic and that was narrowed. I could summarize for you, but Steve already did a good job to a small degree.

I recommend you find article in scolar.google.com on this topic and you will see research by practitioners who itemize. My personal favorites came from studies performed within the PMI (Project Management Institute for those unfamiliar).

The articles were written by Jiang (might be spelled Jieng), Klien, et. al. The et. al is because they had a third party who was different in all the research.

Basically, they looked at communication, collaboration, teamwork, project manager effectiveness, user effectiveness, policy creation, roles, and so on.

The questions you ask are the issues that PM's and IT professionals have been dealing with for a long time. Try to find the OASIG Study (1996) or The Chaos Report (1995). These are foundation studies of WHY IT PROJECTS fail.

by gene.fellner

1. What is the best way to obtain user requirements?

You have to understand each other. The user must have at least a minimal knowledge of IT, and you must have major knowledge of their business. 75% of requirements are not stated. In some cases it's because the user isn't conscious of them, in others it's because they assume since you live in the same country that they don't have to tell you how things work. In either case, they expect you to come to the meeting knowing most of what they need in advance.

Once you've got the requirements, they have to undergo inspection. Users will submit requirements that conflict with each other, don't make sense, or simply don't really specify what they want. Defects in requirements are the most expensive cause of software failure (they usually are discovered in production), the most difficult to catch during testing (after all, what do you test against?) and the most difficult to repair (they require starting over and rebuilding the software).

2. What are the most common reasons for the system not meeting the user requirements?

In my experience, this is covered in my previous answer. The developer doesn't really know what the users want, or the users didn't really ask for what they wanted.

3. How do you handle users who constantly change their mind about what they want?

If it's a serious problem, not just a gripe, you need a more formal project management methodology. The particular activity you lack is known as change control, configuration management, and various other names. It implies that you have a formal, approved project schedule and budget. Every time you receive a request to change the project requirements, you have to modify your project plan and budget to accommodate the extra time and work.

Your users need to understand that changing software once it's under development is almost as much work as changing it when it's in production. They wouldn't expect you to do that for free so they should understand that you can't respond to changes in requirements for free either.

If you don't have a formal software development methodology, this problem is a perfect example of why you should have one. This is the most common and most troublesome problem that project managers encounter. Failure to accommodate requirements changes realistically, by updating the schedule and budget, is one of the biggest causes of project overruns and failures.

If your users just can't help themselves because their requirements honestly do change frequently, then you need a methodology that can accommodate these changes, something more iterative than what you're now using. You may also need a more end-user oriented development environment, for example object-oriented, that allows the users to change their own processing rules without needing your help.

by tony hopkinson

Project management is about coping with
The user doesn't know his requirements.
The analyse didn't completely understand the requirement.
The user's requirements changed and the analyst's understanding of the requirement improved. The longer the project the more this happens.

As soon as you approach an IT project in terms of locking down requirements and expectations, you fail, because you are ignoring the fact that things will change. Everyone has got to stop p1ssing in the wind and start sailing with it.
Change is a given, if your system can't cope with it, it won't work.
How you cope with it is something else entirely.

by softcorp.us

Regarding:

"...As soon as you approach an IT project in terms of locking down requirements and expectations, you fail, because you are ignoring the fact that things will change..."

Strongly disagree. More below.

---

"...Change is a given, if your system can't cope with it, it won't work..."

Of course requirements are going to change. They "always" do. The project management process should expect them and accomodate them.

In a significant IT project, failure to formally document requirements and design, and obtain formal sign-off, produces many problems that would not have occurred otherwise. But, the requirements are not "locked" so tightly that they cannot change. They are "locked" until they must be "unlocked" so they CAN change. The "unlocking" step means everyone knows something significant has occurred and other things may have to change as well, specifically time and cost. Features, time, and cost. Pick any two.

AN ANALOGY
It is well understood that skyscraper design cannot be changed easily in the middle of building the skyscraper. A complex IT project is fundamentally no different. To approach it any less formally is folly. It doesn't matter who misunderstood who. Even in California they don't build skyscrapers "California-style" :-).

by tony hopkinson

about having no controls. You can't manage anything if you don't know where you are, where you want to go, and at least the steps in how to get there. Seeing as all three of those factors could change, a system without considerable flex could be successfully managed but in real terms be a failure, simply because it didn't deliver what was eventually required.

The key to me is not locking down user requirements, designs and methods, it's managing expectation (scope ?).

The big failing we had was QAD in my opinion, because it gave a very false impression of what the expectation was by highlighting the very shiny but extremely empty box that the product came in.

If you can't lock down requirements, and at higher level evolving business sytems that's next to impossible. Iterative development is the way forward, but that shouldn't be by final deliverables but by functionally phased deliverables.

So you do your design on the basis that it is going to change, lock the requirements down in phases, but do not design something that would be very difficult to evolve.

Far too many times I've been called in to sort things out where initial design decisions taken on the basis of 'known' requirements have cut off any potential for evolution of the design at the knees, yet that evolution was obvious to anyone with more than three brain cells.
With the extra two I've got, I've either had to start again or do some unsightly and counter-productive bodge because the people in charge went into bean counter mode.

How else do you explain a customer ordering system design that could only cope with one order per customer for instance, which ended up being one of my unseemly bodges when the product hit it's design limits.

by devrockroy

1. User requirements are something that keep changing. Stated fact, even considering ourselves trying to buy something. Hence, upgrades and so on. The best way is to meet the end user, and help them right the specifications. It would mean more investment in time (and money) but would help identify what they needed then. Consulting at this stage would be to extrapolate (or project) in to the future and think about the areas that they would need to use it. The best way ahead would be to raise limitation issues, so the users are aware of the boundaries of the application, and do not crib about wanting more.

2. Considering that the project is delivered with the features that are required, using the technology promised; the only reason that the user would not be satisfied is because the need/ perception has changed. Taking up from my first point the idea of placing limitations early on would cover this issue since the users would be aware of the limitations and would have to accept them for what they are.

3. You cannot handle that. Users are bound to change their mind, for an evolving product the more they are linked to the development of the product. In times of cost of development, the ideal way would be to partner the user in a manner that the cost of the project is proportional to the time used. That is a method of sharing project management responsibilities. Most clients would be unwilling to pay the extra cost, in which case a time matrix with impact analysis would give them an estimate of where they stand and the losses that would be incurred (financial, if any). Linking time with money generally stops a user from making changes to the end product.

by tony hopkinson

will work as long as the client's expectation is managed properly. When you glean requirements, potential enhancements and evolutions, you have to make it clear to the client that their requirements will change. If they aren't prepared to pay the extra cost both in terms of delivery schedule and cost, then something that they did want can't be provided.
It's that expectation that has to be at the forefront of a development.
Seeing as we in IT are meant to business oriented, why do client's insist that we take a loss because they didn't know/ didn't communicate what they wanted ?
If we screw up of course we pay, if they do, they should.

by softcorp.us

Hello Tony...

Ok. Got your point and agree. If practical and possible, the analyst should avoid designing the project into a "corner" such that it is difficult to change the design to accomodate requirements changes.

Good point.

by apotheon

It goes further than that. Flexibility of design should be worked into it regardless, by way of modularity and universality of intra-project connective bits.

This allows parts to be reused, allows for more efficient design because you don't have to do the same thing with different stuff using reimplementations of the same basic process, and grants greater extensibility and ease of maintenance further down the road. The fact that this approach deals well with specification changes during initial development, thus helping at the project-management end, is just one of the most immediately noticeable effects.

For more on how to design systems that way, I recommend reading a book called [u]The Pragmatic Programmer[/u] by Andrew Hunt and David Thomas, and another called [u]The Art of Unix Programming[/u] by Eric S. Raymond. Raymond's book is useful not just to Unix programmers, but to all programmers: it discusses the philosophy and practice of programming (design, implementation, et cetera) that has arisen in the unix environment. Its origins in unix don't limit that philosophy, though, and its lessons can be applied in any computing environment.

Actually, [u]The Art of Unix Programming[/u] is available, in its entirety, online. A link follows:

http://www.catb.org/~esr/writings/taoup/html/

by the admiral

It has been my expeience that there are more than one way to skin a cat. Let me hit your question point by point.

1. What is the best way to obtain user requirements?

There are several ways. First is to watch what a user does for a period of time (Usually two or three days.) From there you can pull out how they use the application, what common keystrokes they use, and a plethora of other information. It is important to get more than one person using the application to sign up for this, since one person uses the application differently than others. Second, Survey the crowd. Split the survey up into three parts. THe first part of the survey, ask what the old system was, how they used it, what kind of throughput they got, and other information pertenent to the system. The second part of the survey, ask for complaints about the current system, how they use it, and why they think the system is good, bad or horrid. The last part of the survey is to find out what kind of features that you may want to add into the application or thing that you are applying things to. Lastly, you want to do other personal interviews to see WHO is using the application, so you can see what your target audience is. Management may use the system, but the grunts may not, so you want to word the application interface around that information.

2. What are the most common reasons for the system not meeting the user requirements?

Not listening to the user requirements carefully or not caring how to program the system to meet the requirements. Secondly, sometimes the user requirements will not work or are not compatable with the current system, therefore you could not possibly program into the system. In those cases, you have to tell the client that the requirement could not be met and give them a reason, rather than have them come back and claim that the solution was a failure.

3. How do you handle user who constantly change their mind about what they want?

Program everything that they want into the system, but comment it out. When they want something and the module is done and they change their mind, you can also change the code so that remains hidden. Then when they do want it, you can charge them full price for 15 minutes of work 8-). Just kidding. You can then go in and apply changes that will make the system work for the customer without too much problem since it was integrated into the system initially.

by tony hopkinson

That's the way I work. Not so much so I can charge big bucks for very little work in version 3 after all I didn't charge them for all the actual work in version 1.
No, I work that way so there can be a version 3, as opposed to a rebrand and rewrite because I coded my self into a corner in version 2 or 1 as is more often the case.

Just a stubbed out method, can cause you to consider an extra few lines of code, so you can still put something in it's body when you do come to need it.

Good code is code that can be changed/re-used with a minimum cost. If you know what you are doing it only adds to the code base, not the foot print, costs very little and can save you a fortune.

Either you get round the foolish error someone made interpreting a customer requirement, or you keep the customer by swiftly meeting their new and improved requirement.

I see it as a quality requirement, something that's been sadly lacking in IT development for a long time.

When I learnt software development in the old Ivory Tower top down design waterfall approach, it was always accepted that getting the requirements right first time, was cost effective because it avoided re-work. OO and drag and drop in a sensible world should have allowed us to develop higher quality applcations faster, not lower quality ones much faster.
As it was (changing a bit now) instead of getting more milk out of the cow, we got more ****.

by bucky kaufman (mcsd)

1. A: User requirements for what?
2. A: The reason user requirements are most commonly not met is that the scope of the project was never defined.
3. Depends on your relationship to the user.

by tony hopkinson

That is thing thing you want set in stone. Allow a change in scope during a project and you might as well start again, because what you were designing isn't what you will be doing now.

If you fail to recognise a change in requirement as a change in scope, you are ankle deep and head first in deep poo. Amazing how often that happens, he bubbles, wiggling his toes.

This discussion is closed to new replies.