Should every task link to another task?

[Cole]

I have fairly large project plan. Some tasks have no dependencies therefore
no links to other tasks. Should I create links to another task, perhaps a
Start-to-Start, to link it into my project properly? I am worried that when
I start assigning resources these tasks will be set to start earlier than is
necessary.

 

[Jan De Messemaeker]

Hi,

What exactly do you mean by link into the project properly?
When you use the word "properly" I suspect they have some sort of dependency
anyhow.

If they have no dependency at all, then the earliest start is the project
start and the late fiinish is the project finish.
What are the criteria to say it will start "earlier than necessary"?
Isn't "necessary" the expression of some dependency?

In other words, to me a project plan should reflect reality as much as
possible..

Hope this helps

 

[Steve House [Project MVP]]

Dependency links, IMHO, should reflect the flow of deliverables through the
project. They reflect sequences dictated by the nature of the process
process itself - we have to design a product before we build a protype, for
example. So your situation where some tasks don't have predecessors and/or
successors is actually not all that unusual. But it's a good idea to mark
the start of the project with a "Start" milestone and mark the end with an
"Finish" milestone. In that case, we can say every task has at least one
predecessor - if it's not some other task the predecessor is the Start
milestone - and at least one sucessor, the Finish milestone of nothing else.

Tasks really can't start earlier than necessary because you don't want to
schedule them any later than you have to. You always want to get the
project done ASAP and so tasks need to be scheduled as earlier as possible
after the start date rather than as late as possible that meets the
deadilne. If there's some real reason you shouldn't have the task start as
early as Project places it, after assigning resources etc and doing leveling
it's still too early for some reason - perhaps parts aren't available that
early, for example - you can use a Start No Earlier Than constraint to move
it back later in the plan.

 

[davegb]

Hi,

What exactly do you mean by link into the project properly?
When you use the word "properly" I suspect they have some sort of dependency
anyhow.

If they have no dependency at all, then the earliest start is the project
start and the late fiinish is the project finish.
What are the criteria to say it will start "earlier than necessary"?
Isn't "necessary" the expression of some dependency?

In other words, to me a project plan should reflect reality as much as
possible..

Hope this helps

If a task can't start prior to a specific date, it is almost always
because of one of 2 things. The first is an external constraint, like a
delivery to a customer who requires it by that date, or a government
regulation that requires it. The other reason is that it depends on
some other task or tasks. Almost all tasks have one or the other of
these conditions. Exceptions are very rare.

If you think logically about it, if a task has no predecessor and no
successor, and no time constraint on it, it can start the day the
project starts, and can finish the day the project finishes. So it can
be done anytime during the project duration. How many tasks like that
do you have on your projects?

I teach my students to make sure their projects have "Schedule
Continuity". That is, every task, with very few exceptions, has at
least one predecessor and one successor. The only common exception I
can think of, after over 25 yrs of CPM scheduling, are regularly
scheduled meetings. Like a weekly Project Status meeting. It's not
really dependent on anything else happening, it just occurs every week.
So doesn't have to be linked.

To achieve Schedule Continuity, I was taught by a scheduling expert
many years ago to create 2 special milestones, "Start" and "Finish".
Any task that does not otherwise have a predecessor is a successor to
"Start" and any task that does not otherwise have successor is a
predecessor to "Finish". If you look at scheduling software developed
by professional schedulers (Project was not), these 2 milestones are
built in to every schedule, and can't be deleted. You're forced to
create your schedule linked between these 2 milestones. This is the way
the "experts" who developed the software are trying to show that
everything is linked. In this system, even the meetings would be
following the Start milestone and preceding the Finish milestone and
linked accordingly.

There are those who disagree with this methodology. In my experience,
it resolves many of the problems people have in getting Project to
function properly, especially with features like tracking, resource
leveling and others. Some consider it a "best practice", I consider it
a neccessity for a fully fuctional Critical Path Method schedule.

I hope this helps in your world.

 

[Rick Williams]

The short answer is Yes. In the real world, though, an MS Project file is
often created when the project is already underway. In this case only, it is
acceptable to have in-progress tasks that have no predecessors. But all
tasks should have a succesor, with a flow of work that ends at a single
point, i.e. "End of Project". Think about it logically - if a task is not
required for anything else (has no successors), then you are in effect
saying that it is not necessary for the completion of the project, and if
that is the case, why do it at all? Delete it and save money.
Rick Williams

 

[Steve House [Project MVP]]

The planning phase of a project should come well before the inception of
work. How else are you going to know what work needs to be done, what
resources will be required? In the best use scenario, the project plan is a
dynamic planning and forecaasting tool used to create the framework in which
the project will take place, not just documentation. I grant you, some
organizations use informal processes to get a project underway and only
create a more formal project plan when they realize they're heading for
problems without it. But what I suggest in that situation is NOT to just
pick where you are and ignore what came before. IMHO, the best way to get a
project plan online in MS Project when the project is already underway is to
pretend for the moment that work has not started. Set the Project Start
DAte to the date work first began and enter he plan from scratch including
work that has been done as well as work remaining to be done in the future.
For tasks that have been done, enter the durations that they actually
required. Enter Actual Start and Actual Finish dates in the tracking table
just as if you had been tracking progress from the start and bring
everything up to date. Now you'll be on track with a plan that works.

 

[Rick Williams]

I agree absolutely. Unfortunately, sometimes you start with the results of
someone else's mess and have to work with what you have!
Even better, use historical data if it is available to estimate durations
and costs. And ideally, develop optimistic, most likely, and pessimistic
durations and costs. Avoid any fixed dates. Then use a simulation tool, find
all the paths with significant schedule and cost risks, develop contingency
plans for them, and incorporate them into your plan. And finally, update the
project as often as possible!
Rick Williams