none
Hiding contingency time RRS feed

  • Question

  • Hi all,

    I was asked this question today and did not know how to answer it...

    "How do I program contingency time into my project sheet in a way that the client cannot see it?"

    Their issue is that they, like most businesses, schedule contingency time into their project plan to keep things somewhat smooth and predictable.

    However, their client sees contingency time and always wants it taken out so the project can "finish faster".

    My suggestion of talking with the client about it was shot down - apparently this simply does not work for this client.

    My only suggestion was to roll-up the contingency time along with the work into summary bars - but the clients wants to be able to see the full detail ("all the blue bars") in the project sheet to examine them.

    They are not in-depth users of Microsoft Project, but would like to like to keep things as honest as possible in Project's understanding of the real world (they don't do costing at all, but they're aware that adding an extra day or two into tasks is not good for costing or for remembering what the real duration is, etc).

    Suggestions appreciated - I've pulled a blank on this one as my "Be honest to Project and Project will be honest to you" philosophy doesn't achieve what they want it to on this one...!!

    Thanks!

    Wednesday, October 10, 2012 7:58 AM

Answers

  • One option would be to factor the contingency time into the durations/work values at task level.

    So if you've got 10% contingency tacked on the end of a work package at the moment, take it out, and add 10% to each of the tasks within the work package. You might want to document somewhere (i.e. not in the schedule itself) the basis of your estimates +10%.

    That way you've still got the same level of contingency across the project as a whole, but its not apparent from reviewing the schedule itself - you'd need to review the schedule plus the basis of estimate document to get that picture.

    Wednesday, October 10, 2012 8:26 AM
  • There is a section in Eric Uyttewaal's Project 2003 book (Methods to hide the time buffer. p544) It is partially available for preview: http://books.google.com.tr/books?id=pdeeYUwJcaQC&printsec=frontcover#v=onepage&q&f=false

    or fully available in p476-478 of Project 2002 book: http://books.google.com.tr/books?id=31AuYW6TbGUC&printsec=frontcover#v=onepage&q&f=false

    Overestimate lags
    Insert extra holidays in the project calendar 
    Decrease the working hours in the project calendar
    Ignore the benefits of learning curves
    Introduce ramp-up and wind-down factors
    Introduce distraction factors for high-focus tasks
    Introduce extra revision cycles
    Keep maximum units of resources low
    Assign one scarce expert to more tasks than needed
    Create extra inconspicuous tasks
    Pad the duration and work estimates
    Don’t use overtime yet
    Set extra soft dependencies
    Be pessimistic about material delivery dates
    Insert extra holidays for your critical resources

    I believe "padding the estimates" and "using pessimistic material delivery dates" are the most common ones utilized in external projects. 

    Pls post back for further questions.






    Wednesday, October 10, 2012 8:51 AM
  • This is a question about risk. Risk management is a key and explicit area of project management. Applied to estimating, there is a probability distribution of effort/duration values that would apply to every task. Consider the task:

    • Optimistic estimate - 20 hours - say a 10% chance you can achieve this
    • Most likely estimate - 33 hours - a 50% chance you can achieve this
    • Pessimistic estimate - 52 hours - a 90% chance you can achieve this
    • Delphi technique combined estimate - 34 hours (1x20 + 4x33 + 1x52   /   6)

    What do you put in your plan for customer consumption? It depends on how you will be judged. By the time you take into account all the tasks in the plan, there is virtually no chance that you will achieve the optimistic estimate. There is a very good chance (not 100%) that you will achieve the pessimistic estimate.

    There is no shame to acknowledging the uncertainty. The shame should be on the client for expecting certainty. What you can do is describe the risk factors and how your plan is structured to respond to them. At a minimum you should find out what is the most important factor for your client: Earliest possible but accept there is a high probability you won't meet it? Middle of the road estimate with a 50% chance of meeting it? Safe - sure to meet it unless we are really unlucky?

    Check out "The Flaw of Averages" Savage, Markowitz, Danziger;    PMBOK from PMI.

    Wednesday, October 10, 2012 1:18 PM

All replies

  • One option would be to factor the contingency time into the durations/work values at task level.

    So if you've got 10% contingency tacked on the end of a work package at the moment, take it out, and add 10% to each of the tasks within the work package. You might want to document somewhere (i.e. not in the schedule itself) the basis of your estimates +10%.

    That way you've still got the same level of contingency across the project as a whole, but its not apparent from reviewing the schedule itself - you'd need to review the schedule plus the basis of estimate document to get that picture.

    Wednesday, October 10, 2012 8:26 AM
  • There is a section in Eric Uyttewaal's Project 2003 book (Methods to hide the time buffer. p544) It is partially available for preview: http://books.google.com.tr/books?id=pdeeYUwJcaQC&printsec=frontcover#v=onepage&q&f=false

    or fully available in p476-478 of Project 2002 book: http://books.google.com.tr/books?id=31AuYW6TbGUC&printsec=frontcover#v=onepage&q&f=false

    Overestimate lags
    Insert extra holidays in the project calendar 
    Decrease the working hours in the project calendar
    Ignore the benefits of learning curves
    Introduce ramp-up and wind-down factors
    Introduce distraction factors for high-focus tasks
    Introduce extra revision cycles
    Keep maximum units of resources low
    Assign one scarce expert to more tasks than needed
    Create extra inconspicuous tasks
    Pad the duration and work estimates
    Don’t use overtime yet
    Set extra soft dependencies
    Be pessimistic about material delivery dates
    Insert extra holidays for your critical resources

    I believe "padding the estimates" and "using pessimistic material delivery dates" are the most common ones utilized in external projects. 

    Pls post back for further questions.






    Wednesday, October 10, 2012 8:51 AM
  • This is a question about risk. Risk management is a key and explicit area of project management. Applied to estimating, there is a probability distribution of effort/duration values that would apply to every task. Consider the task:

    • Optimistic estimate - 20 hours - say a 10% chance you can achieve this
    • Most likely estimate - 33 hours - a 50% chance you can achieve this
    • Pessimistic estimate - 52 hours - a 90% chance you can achieve this
    • Delphi technique combined estimate - 34 hours (1x20 + 4x33 + 1x52   /   6)

    What do you put in your plan for customer consumption? It depends on how you will be judged. By the time you take into account all the tasks in the plan, there is virtually no chance that you will achieve the optimistic estimate. There is a very good chance (not 100%) that you will achieve the pessimistic estimate.

    There is no shame to acknowledging the uncertainty. The shame should be on the client for expecting certainty. What you can do is describe the risk factors and how your plan is structured to respond to them. At a minimum you should find out what is the most important factor for your client: Earliest possible but accept there is a high probability you won't meet it? Middle of the road estimate with a 50% chance of meeting it? Safe - sure to meet it unless we are really unlucky?

    Check out "The Flaw of Averages" Savage, Markowitz, Danziger;    PMBOK from PMI.

    Wednesday, October 10, 2012 1:18 PM
  • Hi Andrew, Project user and GMcH.

    I just wanted to write and thank you all for three unique and very useful approaches to the issue - I have passed all these on and I'm very thankful for your insights.

    With kind regards,

    Nick

    Sunday, October 14, 2012 9:42 PM