Setting Realistic Goals based on Probability Analysis

There should be a Business Objective (BO) for each and every organization. It can be in terms of profitability, time to market etc. Basically it is chosen on the basis of work being handled by the organization. Here the problem is to map the BO from enterprise level or organizational level to each work units inside the organization. Based on the work units inside the organization, a second layer of objectives might need to be defined (Process Performance Objectives- PPOs, in CMMI terms). The parameter for PPO should be taken based on the critical set of parameters which needs to be monitored. For example to improve profitability, productivity can be critical parameter. So profitability will become the parameter for BO and Productivity as the parameter for PPO. Based on the PPO defined at the organizational level and customer requirements project team needs to come up with project specific goals. For a CMMI high-maturity compliant organization, probability analysis should be there providing enough evidence for achieving the targets. As the PPOs are defined for a range at the organizational level, the probability analysis done at the organizational level won’t match with that at the project level as project goal will be more stringent usually. Hence blindly following the organizational PPOs and corresponding probability analysis might lead to irrelevant conclusions.

Consider a baseline deduced for productivity in an organization as below

Lower control limit= 24 units, Central Value = 27 units and Upper control limit=30 units

There can be four goal statements as given below.

  1. Goal = Baseline
  2. Improve average
  3. Reduce sigma
  4. Improve average and Reduce sigma

 In every case we have an average and sigma. From the organizational perspective range is more important than average (doesn’t mean that average is not important). But for a project which is always concerned about a single value (central value) and not a range, the probability to get more than the central value is only 50 % with the current process. i.e in the case of organization A, current baseline data shows that to get a value >27 units of productivity, the probability is only 50 %.

Analysis of First Case: Goal = Baseline

Goal = Baseline 24 to 30 units

Probability to achieve the targets with the current process performance is 99.97 %. It means no need to trigger further improvement initiatives and the organization can continue with the current process. In such scenarios proper attention need to be paid while goals are defined for the projects. It can have the below cases.

  1. Goal >=24
  2. Goal>=27
  3. Goal>=30
  4. 24<=goal<=30
  5. 24<=goal<=27
  6. 27<=goal<=30
  7. a<=goal<=b ( a–>24 and b–<30)

The problem with the above scenario 1, 2, 3:

  • Any higher value of productivity say like 100 is acceptable for the project manager which in turn means a lack of control in the project execution.

The problem with the scenario 1:

  • This may result in the reduction of productivity ultimately if every project is targeting for the same.

The problem with the above scenario 5, 6, 7 :

  • Probability to achieve the target is 50 % if organizational processes are continued. Then either the risk has to be accepted by the project manager or come up with improvement initiatives to improve the probability of occurrence.

The problem with the scenario 4 :

  • The spread of the data is too high as a PM is concerned to have a better control over the process even though there is 99.7 % confidence of success.

So majority of the organizations focusing CMMI practices fails to map this probability analysis.

There two main situations which we have come across the above 7 scenarios.

  1. There need to be a limit or range of values.
  2. A minimum Probability accepted by the organization to attain the target

In this case organization can decide on the critical probability value with which we can accept productivity, say like if 75 % probability is there to achieve the target, the PM can go ahead with the project defined process. The risk in losing the targets is 25 % and which is acceptable to the organization. So if a range is defined as ‘average – 1 sigma’ to ‘average + 3 sigma’, the probability will become 83. 85 % (=49.85 +34). So the project can have a goal as 26 to 30 with a probability of ~84 %.  With Process performance Models (PPM) project should get this probability after simulation.

Analysis of Second Case: Improve average

Goal = 30 (LSL) to 36 (USL) units.

Currently probability to achieve the above target is 0 %. Definitely organization needs to come up with improvement initiatives. For example, if it is observed (in the industry or within the organization ) that a particular tool usage can enhance coding activity, especially the coding speed which in turn may results in an increased productivity based on the correlation between productivity and coding speed. First of all after choosing an improvement initiative, an organization needs to do the CBA quantitatively to decide the Return on Investment on the tool usage. There after correlation analysis should be done between the parameter identified for improvement initiative (say it as x) and PPO (say it as y). Then a PPM shall be built between x and y. In the PPM there can have other parameters which are affecting PPO. In the PPM sometimes the parameter x may be already in practice as it was already identified as critical sub process parameter affecting PPO.  Current range of sub process parameters used in PPM shows the current process performance and hence with the current PPM the probability to get between 24 to 30 units of productivity is 99.97 % and to get 30 to 36 is 0%. Now if the limit of coding speed is changed to the anticipated range which will obtain after implementing the improvement initiatives, the probability should get as 99.7 % for the range of 30 to 36. In that case a project can target for 32 to 36 with a probability of 75 %. If the 99.97 % probability is not achieved by organization for the targets, further improvement initiatives are required and the steps need to be repeated.

 In the same manner third and fourth Case can be analysed to achieve the targets (average – 1 sigma to average + 3 sigma) with a probability of 75 %


The author is a Quality Assurance professional by experience. Part Quantitative data analyst, part consultant for quality and information security practices, part software tester, she is a writer by passion and blogs at and

You may also like...

Leave a Reply