Page 1 of 1

Best Practice - Timelines and percent split

Posted: Mon 07 Mar 2011 11:27 am
by kershawp
Morning all

Newbie here so please be nice to me! ;-)

Does anybody know if there is a "Best Practice" split between time spent in development against the rest of the project lifecycle. i.e. if the development is taking 8 weeks, what should the rest of the lifecycle take (should testing be a prescribed time etc), time wise if based on best practice. We are following a typical waterfall methdology.

Thanks in advance for any help here.

P ;-)

Re: Best Practice - Timelines and percent split

Posted: Mon 07 Mar 2011 10:46 pm
by dhaughey
The Mythical Man-Month: Essays on Software Engineering, by Frederick P. Brooks, gives a rule of thumb for scheduling a software task:
  • 1/3 Planning.
    1/6 Coding.
    1/4 Component test and early system test.
    1/4 System test with all components done.
That's 33% spent on planning (before writing any code), 17% on code writing, and 50% on testing, which Brooks says is always underestimated.

One of the quotes I've always remembered from the book is, "Assigning more programmers to a project running behind schedule, may make it even more late." Keep this in mind when your customer asks for an earlier delivery and thinks throwing more resource at the problem is the answer.

Other insights from Brooks are:
  1. Estimation techniques are poorly developed.
  2. We tend to confuse efforts with progress.
  3. We are uncertain about our estimates.
  4. Schedule progress is poorly monitored.
  5. In case of schedule slippage, the natural response is to add manpower.
A must read for anyone involved in developing software.


Re: Best Practice - Timelines and percent split

Posted: Tue 08 Mar 2011 2:50 am
by satisfactionuk
Hmmm, I usually find that there are two trains of thought depending upon which side of the person’s brain is dominant. On the one side you have the mathematically ordered administrative type of person who loves to write list and has lots of rules for just about anything and on the other side you have highly creative visual people who hate rules and formulas.

Of course the prescription expounded above by dhaughey more than adequately answered your question and I am definitely not making any assertions as to which side of his brain he was using. Lol

However, I do believe that in all fairness your question just needs common sense and logic applied to it.

Baring in mind the size, complexity, time and cost constraints and the availability and skill of the resources every project and phase there in will take as long as it takes to complete.

I am sorry to disagree with the formula provided by the author of that book dhaughey but logically speaking, you could give the same company exactly the same project to complete at two different times and come up with two entirely different time scales.

The reason for this is because at the beginning of the year you might have a highly professional expert who is motivated and competent at the project or task at hand and three or six months later s/he leaves and the replacement might by significantly less motivated or less competent.

However as your fully aware, in project management there is a myriad of other variables that can derail even the best-laid plans.

So to answer your question, there are no set rules or formulas for how long a project or section of a project should take just professional guidelines for best practice for getting the job done.
Kind regards

Stephan Toth

Re: Best Practice - Timelines and percent split

Posted: Tue 08 Mar 2011 11:11 am
by kershawp
Thanks for the replies all, very helpful.

I do know that the best laid plans and all.......;-) It was more around the planning and trying (for our many repetitive projects) to put some indicative framework in. I have also found another formula from PMBOK which, I think, states the 60, 10, 30 rule following Detailed Requirements, i.e. 60% planning, 10% coding and 30% testing. Applying this does not come out too far from the rules here.

Thanks again all.


Re: Best Practice - Timelines and percent split

Posted: Wed 09 Mar 2011 1:00 am
by satisfactionuk
Hi, you didn’t say that you had several projects already completed.

In this case I would open a spreadsheet and using cells A1:C3 put in the headings Planning, Coding and Testing.

The next stage would be to take the actual times from each of your completed projects one at a time and insert the values under the relative headings.

The next stage would be to sum up each column to find the total and then divide by the number of entries to get the average.

The next stage would be to sum up the averages and place that value in cell DX

Finally its a simple operation to determine the percentages by typing in the cell under each of the cells containing the averages =SUM(Ax/Dx*100) (where x equals the cell number) and so on for each of the other two cells containing the averages.

Doing this will give you a much more precise ratio that is determined by actual figures supplied by projects that your firm has completed in the past.

Of course if you want to take this a stage further you can then run a Monte carol or Latin Hypercube simulation using either @Risk or Crystal Ball to give you a three point estimate on all of the totals.

Doing this (say 1000 iterations) with a plus or minus of say 10 or 15 percent will aid you to be able to say that the mean planning time is Y time however if all goes well we can do it in X time and if we run into problems then it will take Z time. And of course the same for Coding and Testing.

As you do more projects you can add the new figures into the spreadsheet and repeat the process and refine your percentages according to real life data.

Surely this would be more meaningful to your company and far better than taking some arbitrary formulas from a book

I hope this helps you.

Kind Regards

Stephan Toth

Re: Best Practice - Timelines and percent split

Posted: Wed 09 Mar 2011 3:13 pm
by dhaughey

I like the whole spreadsheet idea. Much better to have data based on your own firms experience than formulas from academia.

Using the Monte Carlo method of estimating can give a result close to the actual time it will take for each activity. It's based on the generation of multiple trials to determine the expected value of a random variable. Although there are commercial packages to do this for you, you can use a spreadsheet to run simulations. Take a look at this guide:

Monte Carlo Simulation in MS Excel

(Just substitute cost with time)


Re: Best Practice - Timelines and percent split

Posted: Thu 10 Mar 2011 12:22 am
by satisfactionuk
Hi, Wow lots of maths lol, though I did quantative maths several times at college and uni (and passed) I hated it lol, I would much rather let a computer do the hard work. Don’t get me wrong, I still study statistics and managerial economics but I translate the problems onto my computer rather than do them by hand. Guess I am more inclined to use the creative side of the brain.

Anyway, the reason for this response is a nice little freebie called MicrOsiris in my opinion its not as intuitive as @Risk or Crystal Ball and not as easy to use but with some practice it is well worth downloading.


I copied this from their information.

Management Software (Freeware)
Version 14 2/5/2011
MicrOsiris is a comprehensive statistical and data management package for Windows developed by Van Eck Computer Consulting and freely distributed. Derived from OSIRIS IV, a statistical and data management package developed and used at the University of Michigan, MicrOsiris includes special techniques for data mining (SEARCH) and analysis of nominal- and ordinal-scaled data (MNA, MCA) and an interface to IVEware.

The MicrOsiris IVEware interface command, IVEWARE, invokes the Srcware version of IVEware (installed with MicrOsiris) that can:
  1. Perform single or multiple imputations of missing values using the Sequential Regression Imputation Method.
  2. Perform a variety of descriptive and model based analyses accounting for such complex design features as clustering, stratification and weighting.
  3. Perform multiple imputation analyses for both descriptive and model-based survey statistics.
  4. Returns an imputed data set for further analysis in MicrOsiris.
MicrOsiris accepts data from SPSS, SAS, STATA, and Excel as well as from other sources, including UNESCO IDAMS datasets and older OSIRIS datasets from ICPSR.

It will run under a Windows virtual machine on a Mac and on Ubunto/Linux under Wine.

"I found that it was the only desktop product that could reference over 800 vars and handle a record > 6,000 chars. Saved my bacon!" from a steady user, 2009
Click here for key benefits and more details
Download MicrOsiris
Free Statistical Software with Reviews
Free Resources for Program Evaluation and Social Research Methods
Free Statistical Tools

Calculation of Statistics in Higher Education
The Decision Tree for Statistics (Freeware; included with MicrOsiris or use link below).

The Decision Tree helps select statistics or statistical techniques appropriate for the purpose and conditions of a particular analysis and to select the MicrOsiris commands, which produce them or find the corresponding SPSS and SAS commands.


Hope this is useful to you guys.

Kind regards

Stephan Toth

Re: Best Practice - Timelines and percent split

Posted: Fri 14 Sep 2018 1:33 pm
by millan123ta
The % Complete fields contain the current status of a task, expressed as the percentage of the task's duration that has been completed. You can enter percent complete, or you can have Project calculate it for you based on actual duration.